Proceedings of the 11th Joint International Computer Conference

CONFERENCE ORGANIZATION

PROGRAM COMMITTEE

PREFACE

KEYNOTE AND INVITED SPEAKERS

CONTENTS

In order to evaluate the performance of AQM (Active Queue Management) algorithms, we present an analytic model based on an extended GI/M/1/N queueing system and the self-similar traffic of the Internet. Using this model, we analyzed the performance of three AQM algorithms (TD, RED and GRED). The analytic results are consistent with those obtained from simulations or experiments reported by other authors.

Practical complex networks in large scale usually bear two major features of “Small-World” and “Scale-Free”. There are limitations if the classical random graph theory is used to explain complex networks. So Degree Distribution, Average Path Length and Clustering Coefficient are introduced to describe the orderliness of complex networks. However, people are still not sure about the basic mathematical theory of small-world and scale-free networks. The significant features of complex networks are studied and simulated in this article. For the complex networks in real world, especially for the next-generation Internet, the influence to the complex networks topology from the major factors of distance, flow and bandwidth, etc. should be taken into consideration as well as nodes and edges. Data Field Method is used to take an attempt in this field in this article.

The “fuzzy” electrical engineering method to online algorithms is defined not only by the evaluation of XML, but also by the theoretical need for reinforcement learning [1,2]. In fact, few scholars would disagree with the development of Moore’s Law, which embodies the structured principles of software engineering. We introduce new multimodal models, which we call AbsorptLydine.

In MPLS networks, in order to optimize performance, some LSPs which are established but not on their shortest paths should be rerouted to their shortest paths. For adjusting these LSPs as many as possible, the reroute sequence should be planned. In this paper, RSP (Rerouting Sequence Problem) are introduced, and relative algorithms are analyzed and compared. On the basis of the algorithms, we propose a new algorithm: MD (maximal decreasing), which can choose a LSP decreasing the most bandwidth occupation in the whole network. The analysis result demonstrates that MD significantly reduces computational complexity and bandwidth change in comparison with previous algorithms.

With the increasing of network applications, routers must perform packet classification at high speeds to efficiently implement functions such as firewalls and provision of QoS etc. In this paper, we analyze the applications of the Hash algorithm in IP classification of multiple fields. And we proposed a novel IP classification algorithm based on the hash_tree structure. The algorithm can make use of the fast search of Hash algorithm, and can process IP classification with the Tbps links speed. At the same time, the algorithm has the characteristics of supporting big rule set and incremental update. It is suitable for flow matching of small granularity such as differentiated services, traffic billing and so on.

Packet-pair has been used as one of the primary means to measure network bottleneck bandwidth or capacity. Yet, most prior proposal tools are sensitive to network status and perform poorly in heavy-loaded network. We present a novel filtering mechanism to address the negative effects of cross traffic. After split the origin set of probe pairs into two packet sets composed of the first and second packet of all pairs respectively, we select the packets with minimum one-way delay in each set and use them to reconstruct a new pair free from interference of cross traffic, from which the final capacity estimates are derived. We show the mechanism in detail and validate it in simulations as well as Internet experiments. Preliminary results show that the proposed mechanism is feasible and robust for the heavy-loaded network, which can produce accurate estimates with relatively fewer overheads compared to similar tools.

In this paper, we analyze the Management mechanism In InfiniBand Architecture. We describe subnet management and general management service in three aspects, such as management model, management entity and realization mechanism.

Anycast is a kind of new service so the study on it is still on the primary stage and inevitably there exist many problems in fulfilling anycast service, one of which is scalability. This paper proposes a scalable architecture for global IP-anycast service on the basis of BGP and ICMPv6 in IPV6 simulation, and analyzes and discusses the feasibility and validity of this architecture, and gives a detail description of its design and implementation. According to the IPv6 simulation, this paper further discusses its performance and error tolerance.

The laboratory is the basic unit of a university. Understanding the traffic of the laboratory network is helpful for management and maintenance of a campus network. Though there are many papers discussing the traffic of campus networks in other nations, few researches have been done on traffic of campus networks in China. With Windump and Ethereal, we have captured the network traffic of nearly 12 giga bytes from a laboratory network of Xi’an University of Engineering Science and Technology in China. And we developed a software tool with VC++ under Windows 2000 professional, which can transfer the captured traffic into a database and make analyses and statistics on it. After analyses of the captured traffic, we understand the traffic distribution over protocols. HTTP accounts for about 73% of the total traffic. FTP accounts for about 24%. Email accounts for about 2%. In addition, we found the traffic is very bursty, it builds up suddenly to thousands times as high as usual. The request sizes of HTTP have a peak around 250 bytes. The mean of response file sizes of HTTP ranged from 8-10 KB. More than 85% responses have a response time less than 0.05 seconds.

Optimal sequences for scheduling divisible load on heterogeneous systems are considered. The communication mode is of non-blocking message receiving. The processors may have different processing speeds, links may have different communication speeds, and both processors and communication links may have different start-up costs. The influence of start-up costs on the optimal processing time is analyzed; and for bounded number of processors with large-scale workload, optimal sequences for distributing divisible load are derived analytic. Numerical examples are presented to illustrate the analysis.

It is important for an interconnection network to route data among nodes if there are fault nodes in the network. Alternative paths in case of node failures can not only avoid communication bottlenecks, but increase the efficiency of message transmission. In this paper, we give an algorithm which, given at most n−2 faulty nodes s and t in the (n,k)-star graph S_n,k, finds a fault-free path s→t of length at most D(S_n−1,k−1)+6 in O(n) time, where D(S_n−1,k−1) is the diameter of S_n−1,k−1. Using this algorithm as a subroutine, we present another algorithm, given at most n−1 faulty nodes such that the result graph is also connected, finds a fault-free path s→t of length at most D(S_n−1,k−1)+8 in O(n) time. The algorithms are optimal in the sense that both the upper bounds in the length of s→t and the time complexity are optimal.

FAST TCP uses queuing delay, in addition to packet loss, as congestion signal but the overestimate of the round trip propagation delay will cause unfairness. In this paper, we proposes an algorithm to better estimate the queuing delay in FAST TCP for the case of a single bottleneck model, so that we can avoid unfairness of FAST TCP. We also made some simulation, and the simulation results show that the algorithm can improve fairness.

This paper considers the analysis on network data flow of discrete-time queuing system. The system consists of single buffer and there serverrs with the same service function and different service rate. In the process of data transmission, when many input terminals send data to a middle note synchronously the waiting data have to queue in the buffer. We use fuzzy control theory to select suitable routes for the data so as to reduce the sojourn time and avoid network congestion. We give different simulations in terms of different conditions. Numerical examples are worked out which show the validity of the fuzzy controller.

The term usage control (UCON) is a generation of access control to cover authorizations, obligations and conditions. The core properties of UCON are decision continuity and attribute mutability, which depend on the attributes of subjects, objects and system. We argue UCON is an attribute-based access control taking usage as a natural attribute handling process. To describe this feature, we identify the taxonomy of attributes, such as mutable or immutable; subject-depended, object-depended, or system-depended; authorization-related, obligation-related, or condition-related; initial or driven. At the same time, we provide the Using model with respect to attribute mutability, which is an abstract attribute-based UCON clearly since it takes usage as direct attribute handling process. Consequently, a redefined UCON, based on the Using model, is proposed to highlight attributes’ importance and to offer an abstract description mechanism that can specify access control within different situations of sixteen sub-models of UCON_ABC model. The most abstract model, attribute-based decision system, is proposed as a final abstract UCON model, which indicates the main contributions of this paper: (1) UCON is an attribute-based access control; (2) decision factors, such as authorizations, obligations, conditions are all attribute-oriented.

At present, capacity is the prevailing paradigm for covert channels and most of current researches are based on Shannon’s channel theory. With respect to steganography, however, capacity is insufficient. Ira S. Moskowitz et al propose “capability” paradigm in [1], but they had not given a feasible algorithm for “capability” description. In this paper, a new capability description for audio information hiding is proposed and a feasible mathematic method is given to determine how much information can be hiding. The proposed capability description can be used to give a prediction framework of steganographic communication in our speech steganography system.

Separation of duty is an important constraint in role-based access control model and is used to resolve interest conflicts among roles. Enhanced separation of duty (ESoD) defines not only SoD on roles but also SoDs on permissions, operations, objects and even users. Further more, ESoD defines conflict transfers in RBAC model elements and role hierarchy which constructed relationships among above isolated SoDs. We comprehensively and formally describe above concepts in ESoD.

Based on the multi-sensor data fusion technology, a new Intrusion Detection System Fusion Model-IDSFM is presented. This model is characterized by correlating and merging alerts of different types of IDSs, generating the measures of the security situation, and thus constituting the evidence. Current security situation of network is estimated by applying the D-S Evidence Theory, and some IDSs in the network are dynamically adjusted to strengthen the detection of the data, which relate to the attack attempt. Consequently, the false positive rate and the false negative rate are effectively reduced, and the detection efficiency of IDS is accordingly improved.

Public key encryption was first introduced by Diffie and Hellman in 1976, but chaos-based public key encryption became a new topic in recent years. In this paper, we design a new chaos-based public key encryption scheme. By using the GNU MP arithmetic library in CYGWIN, a Linux-like environment for Windows, we verify that our scheme is practicable and efficient. On the other hand, cryptanalysis shows that the proposed scheme is secure.

The system call sequences have already become one of the important data sources in host-based intrusion detection system. There are some merits which are high in accuracy, low in false fault and good in stability and so on by using the system calls analysis to judge the intrusion. This paper puts forward a high-efficient and low-loading abnormal detecting method aiming at sequences. The method is based on rough set and capable of extracting detection rules with the minimum size to form a normal behavior model from the record of sequences generated during the normal execution of process. Compared with other methods, the merits of using the Rough set theory to create the normal model are as follows: it is simple to get the training data; the small rules set is suitable to real-time detection, and the process’ abnormal running state can be detected out effectively. Experiment results show that the efficiency of the method in this paper is obvious higher than other methods.

Security architecture model is an intractable research problem in web application systems. In the traditional three-layer architecture model, there is no specific structure and safeguard mechanism for the overall safety of the system. This paper analyzes the security problems of existing web application systems and proposes a new four-layer architecture model for secure web application systems. The four layers of this new architecture are: presentation layer, business layer, data layer and security control layer. Access control and communication security are implemented in our four-layer architecture by adding new functions on top of existing ones to the first three layers, i.e., presentation layer, business layer, and data layer. The new security control layer is mainly used to provide safety services, such as two-way certification of end to end communication, secrecy and integrity of communication data, undeniable service of end to end communication, algorithm of digital signature and security key distribution etc. This paper also defines functions for each layer and describes authentication, encryption, decryption and transmitting data flow for the new architecture. The deficiencies and security problems in the traditional three-layer architecture are resolved in the safeguard mechanism in our new four-layer architecture; moreover this four-layer architecture has preferable universality and opening performance.

The GSHIDS scheme is the only HIDS scheme at present, however, it is not very efficient. In this paper, an efficient hierarchical ID-based signature scheme, where all efficiency parameters are independent of the hierarchy depth, is presented. We solve the problem that signature expansion factor and complexity of verification grow linearly with the number of hierarchy level in GSHIDS scheme. Signature in our scheme consists of just three group elements. In addition, verification only requires two paring operations and one multiplicative operation. Other algorithms in our scheme are nearly as efficient as those in GSHIDS scheme. Such high efficiency is thanks to the new technique adopted in extraction algorithm. We prove that our scheme is secure against adaptively chosen message and ID-tuple attack in random oracle model when GSHIDS scheme is secure. Our scheme can be extended to many aspects and constructed to other efficient schemes such as distributed PKGs scheme, hierarchical ID-based threshold signature scheme, and very efficient forward secure signature scheme.

In 1998, Lee and Chang proposed a group signature scheme based on the discrete logarithm problem. Later, the Ao-Chen-Bai group signature scheme was brought out as an improved scheme of Lee-Chang, however it was proven that could not resist forgery attack. To solve this problem, a new group signature scheme is proposed in this paper. This new scheme extends the algorithm in the Galois Field GF(p) of the original group signature scheme to the ring Z_n, and at the same time, by modifying the joining and identifying stages, so it increases the security performance and the computation efficiency. This scheme can be applied to other ring-based cryptographies schemes.

Concerning the limitation of the attack of Sun and Hsieh to the strong proxy signature scheme with proxy signer privacy protection proposed by Shum and Wei, this paper discovers a more terrible attack to Shum and Wei’s scheme, under which the original signer can forge any proxy signers to generate a valid proxy signature. Based on the discovery, we propose an improvement over Shum and Wei’s scheme to avoid this attack. The improved scheme can resist not only the Sun and Hsieh’s attack but also the serious attack which we have discovered. And the new scheme keeps the merits of the original scheme, with the improved security and practicability.

As web technology has evolved, more and more enterprise applications become web-enabled. In order to secure web systems, a solution based on SMS (Short Message Service) for identifying users is put forward in this paper. As an additional identification of web application systems, especially sensitive web application systems, its implementation and advantage are discussed deeply. At the same time, a special case study, which this solution is applied to XOCECO Enterprise Portal, is given in detail.

Triple DES is the most widely use symmetric cryptographic algorithm. It has many applications in communication and financial area. Operations in triple DES algorithm, such as: permutation, circular shifts, and S-box transformations are time consuming and can not be efficiently processed by traditional CPUs. In this essay, we have explored the tradeoffs in designing a triple DES reconfigurable processor. The optimization techniques we have explored include: parallel processing, loop unfolding, and pipelining. We have evaluated our design on the Xilinx XC2V1000 FPGA platforms and get a throughput of 6973.2Mb/s for triple DES operation.

Mobile agent is a program that can autonomously migrate from a host to another and it provides a useful framework for electronic commerce. But, despite its many practical benefits, mobile agent technology results in significant new security threats from malicious agents and hosts. So, there are lots of works in the mobile agent’s security. Recently Seung-Hyun Seo and Sang-Ho Lee proposed a new secure mobile agent system based on the multi-signature. However we found their protocol can’t provide non-repudiation service, that is an important security criterion in mobile agent system. To overcome this problem, we improved their scheme with undeniable identity based signature. In this paper it is shown that for the nature of identity based cryptosystem (IBC), IBC is more suitable to be applied to construct security infrastructure for mobile agent system than traditional PKI (public key infrastructure) system.

The design of the IP protocol makes it difficult to reliably identify the originator of an IP packet. attackers often use incorrect, or spoofed, source IP addresses. In this paper, We propose a new method for IP traceback that collects audit trails for traffic within the network, and can trace the origin of a single IP packet delivered by the network in the recent past. We demonstrate that the method is effective. We present both analytic and simulation results showing the method’s effectiveness by contrasting with SPIE[1] and Scheme J[2]. We believe that the key contribution of our method is to demonstrate that constructing attack graph is not necessary for tracing.

Active worms spread in an antomated fashion and can flood the Internet in a very short time. Modeling the spread of active worms can help us understand how to monitor and defend against the propagation of active worms effectively. In this paper, we present an active worm propagation model with discrete time, referred to as Discrete Time Worm Propagation (DTWP), which considers the propagation of anti-worms. The figure of simulation by matlab indicates that compared with the AAWP model, our DTWP model can characterize the latter propagation of active worms more effectively.

Traditional access control of CSCW system is lack of delegation mechanism. For improving it and achieving resource share in the heterogeneous distributed environment, Role-Activity Based Delegation Model for CSCW (abbr. RABDM for CSCW) was presented based on RBAC policies and the important properties of activity, object-oriented visual modeling was also further represented using Unified Modeling Language. The model with time and constraint characters solves the issue of authorization centralization in the actual application of Graphics and Images Collaboration Processing System Based on CSCW, and realizes distributed and dynamic characters. Visual modeling also shortens the gap between the abstract theory model and implementation of the applied system, and benefits designing and development of systems based on the self-contained model.

Today, networks are becoming more complex and consequently more vulnerable to various kinds of attacks. So management of networks requires more smart and sophisticated methods. Smart agent technology provides a powerful method for the modeling and development of complex networks. So, we propose an smart agent-based dynamic information security management method to adapt to networks’ vulnerability and complexity, In this paper, we describe our method of dynamic information security policy of networks, which is based on security policies, security events and security concepts. And we proposed a new generation of flexible and adaptable systems for efficient dynamically security management. The smart agent-based approach is innovative because it makes dynamic information security management system to adapt to unpredicted complex evolution of both network environments and intrusions. The implemented system enables the detection of well-known attacks.

The known solutions to achieve confidentiality and authentication simultaneously fail to provide verifiability using standard elliptic curve signature [10][1]. An elliptic curve based signcryption scheme SC-ECDSA was proposed. A one time padding cipher with a session key generated by respective secret knowledge was constructed to encrypt messages. The standardized signature scheme ECDSA was used to sign and verify. Proof shows that SC-ECDSA matches the three security notions: unforgeability, non-repudiation and confidentiality (provable CUF-CPA). For typical security parameters, SC-ECDSA saves 80% computation time and 4.7% message expansion than other schemes. It presents a 29.2% reduction in computation time and a 6.9% reduction in message expansion than traditional Sign-then-Encrypt.

Effective security assessment is important to protect network security. This paper presents an information fusion model based on multi-agent to improve the process of network security situation assessment. The paper discusses the advantages of information fusion for network security assessment. It presents that agent is suitable to implement information fusion model in distributed network system. Then it describes the architecture of agent and formalizes the information fusion model composed of multi-agent. The model comprises four modules: Detection, Classification, Decision and Response. Agents in modules complete different works with cooperation. They automatically collect data about security from distributed network security equipments. The model classifies these data with certain rules and makes decision to reflect security situation. The model proposed in this paper constitutes a basis for information fusion system to implement network security situation.

In 1995, Wu proposed a remote login authentication scheme based on geometric approach without any password table. Wu’s scheme is humanization, simplicity, efficiency and beneficial to smart card. However, Chien, Jan and Tseng presented a cryptanalysis of Wu’s scheme to show that it is not secure. A modified version of Wu’s scheme was also proposed by them. In this paper, we shall point out that the modified remote login authentication scheme is still susceptible to password guessing attack. At the same time, we shall propose a solution to resist such attack.

In this paper, a moment estimation method of the embedding strength for the digital image watermark is discussed. Under the assumptions that watermark are a known, zero-mean, binary valued sequence and that the original image data are an i.i.d. random sequence from a Weibull distribution, embedding strength should be estimated by observing to the averages of the watermarked image, the averages of the original image, and calculating briefly between them. The approach is based on moment estimation method of the watermark embedding strength. The experimental results show that this method is effective for De Rosa et.al.’s watermark detector. Moreover, according to the addition, multiplication and fused embedding rules, we also compare and analyze their performance using PSNR. The conclusions are useful to the research in the future.

Disaster recovery planning is very crucial for highly security-critical information systems to control the disaster recovery activities. Using optimization theory, we present a disaster recovery planning mathematical model in this paper. This model uses mathematical method to express different entities in disaster recovery system such as applications, facilities, resources, sub plans, budget, etc. Through identifying the significance of various applications and expressing relationships among applications, facilities, resources and sub disaster recovery plans, this model can help to select the best set of sub disaster recovery plans. Because of possessing less parameter, the model can be easily implemented. This model has less subjective parameter and can evaluate disaster recovery planning externally and quantitatively.

Access control plays an important role in ERP system. Conventional ways of access control are access control list, access control matrix, and capability listing, which have a rigid security mechanism and lead to poor performance. The paper addresses a RBAC-based user authorization mechanism, which has a layered architecture, including database layer, authorization information layer, authentication layer, and presentation layer. Based on RBAC, security issues are considered at each layer, so that it offers better flexibility and scalability. MVC, middleware techniques are taken to build this security mechanism. According to the hierarchy, it presents the design and implementation of each layer. The system security is enhanced significantly by means of the mechanism.

To overcome the limitation of existing signature-based intrusion detection system in detecting web server attacks, we detect attacks of Web server based on the root causes and the locations where those attacks occur, therefore propose Web IDS model in the paper. The resource monitoring of Web server is the core of this model. It is realized by coordination and cooperation of many resource monitor agents. We Propose an Algorithm of file access analysis and an Algorithm of port information extracting based on protocol analysis for defending of Web Server attack and improving detecting efficiency.

The problem of software generation of the Generalized Self-Shrinking (GSS) sequence is considered. An efficient algorithm with low space complexity is proposed, which takes advantage of the special structure of the GSS. Several methods are discussed and a new block-oriented method is introduced. Then, a two-sequences-combined model is explored to make the algorithm more efficient. A comparison of the experiment result shows that it is competitive with other stream ciphers.

With the constant development of Internet, the security problem of computer network is becoming more and more prominent increasingly. The traditional intrusion detection systems cannot detect the unknown intrusion attack. So, in this paper, the computer intrusion detection system based on immune principle is introduced at first, and then we analyze the methods of computer intrusion detection and present the disadvantages of the original negative selection algorithm. Furthermore, an improved intrusion detection algorithm has been proposed based on the original one. Finally, we contrast the improved algorithm with the original one from time expenditure through a concrete example and it proved that the improved algorithm was much better than the original one.

Web services composition is an error-prone task and involves many aspects such as flow control, data exchange and communication between peers. All these make behavior analysis and verification of it significantly complicated. In this paper, we propose a new technique of composition analysis and verification by using CPNs, which is an extended version of Petri Nets, which have a sound mathematical semantics and a number of existing analysis tools. We provide transformation schemes of composition language into CPNs and a technique to analyze and verify effectively the nets to investigate several behavioral properties. Our transformation technique is essentially independent of the language describing composition. As an example, to show the effectiveness of our technique, we pick up BPEL and translate a specification written in it into CPNs.

To enable the UDDI Business Registry to match web services according to the semantics of service description and service requirement, a method employing ontology to extend the capability of the UDDI Business Registry is proposed. In particular, this paper describes the design of the ontology service built as a standard web service to enhance the flexibility of the method. The mechanism of semantic-enabled matching service based on ontology is introduced. An example of web service matching based on domain ontology is provided. A design of medical diagnosis domain ontology is introduced. The simulation involved on medical diagnosis service based on JADE™, a Java-compliant mobile agent platform from TILab, demonstrated the feasibility and flexibility of this method.

Wireless middleware is a software component that facilitates the interaction of components in the wireless environment. Another popular technology used for distributed communication is XML. XML messaging and Web Services have been used for enterprise integration in the traditional computing environment but its verbose nature does not suite characteristics of the wireless environment. This paper introduces wireless middleware concept, feature and function and discusses its benefits. Then analyzes the wireless network architecture and depicts a middleware solution that uses XML messaging for communication. The use of XML as a possible intermediate data representation for communication was presented and the advantages of using XML were discussed. Finally we give the solution for the wireless middleware communication.

In peer-to-peer based information grid environment, applying ontology to describe resource has become current research trend. However, different peer maybe use different ontology to describe resource metadata according to its domain knowledge. How to discover and match these resources is a desiderate research problem in this environment. In this paper, based on formal description of node ontology and metadata model, a hybrid resource matching approach is proposed. It combines exact matching and fuzzy matching. At the same time, it combines semantic matching and resource instance feature matching. This approach will improve flexibility and accuracy of resource discovery in decentralized environment.

In grid computing environment, applications are frequently executed at sites that are distant from their data. So data access mechanisms that could hide the complexity of remote data accessing via standard I/O interfaces and permit randomly access to any part of a data are required. To address the requirement, a data access service called Random Access File (RAF) service will be promoted in this paper. This service makes use of the Globus Toolkit and provides a set of WSRF service interfaces that can be easily integrated with client applications written by any programming language. With its random access feature, applications can access arbitrary part of a dataset. Transfer data using attachments, parallel data transfer, and buffer adjustment technologies are also used to enhance the performance of the data movement. Finally, this paper presents micro-benchmark studies to provide analysis of the performance improvement.

Compared to today’s traditional distributed systems, the mobile computing environment is very different in many respects. Limited battery life and high communication prices are bottlenecks in mobile computing, so a new method is urgently needed to resolve such problems. This paper proposed a new framework to meet the demands of future wireless applications. In the framework we propose Asynchronous Access Pattern (AAP) and Service Configure File (SCF) supporting disconnection (include voluntarily and involuntarily) operations for mobile clients. AAP allow a mobile user voluntarily disconnect itself from the server to save its battery life and avoid high communication prices. And new wireless applications are easy to be developed/deployed based on SCF in our framework.

StAX is presented by JCP JSR-173 specification which supports XML pull parsing and gives more parsing control to users. This paper describes the design of pushdown automaton model for XML syntax analysis and illustrates the implementation of StAX parser OnceStAXParser. After passing the rigorous XML conformance tests and StAX API conformance tests, OnceStAXParser is optimized from many aspects such as the implementation of pushdown automaton, arranged pre-allocation and lazy processing, appropriate encapsulation strategy, etc. The performance test results from XML Test suite show that the throughput of OnceStAXParser is 28% more than that of BEA StAX RI on the average.

It is an important problem to implement realistic and trustworthy models in Ad hoc networks simulation, and node’s mobility pattern and traffic pattern are two key models. After introduction of obstacle in a simulation terrain, the node’s movement pattern will be affected by collision between nodes and obstacles (static obstacle and dynamic obstacle). The node’s movement pattern can be predicted and decided by using obstacle collision detection method, and obstacles also affect node’s signal propagation. This paper considers above two factors and implements an Ad hoc network mobility model. Simulation shows that this model has a significant impact on performance of protocols.

This research aims to discover the way to adjust the web page hyperlink automatically. First we utilize a WUM (Web Usage Mining) analysis tool to analyze the log file of the website, apply the Greedy Algorithm to the fetched data (or pattern) to extract the most optimum part of the web page hyperlink, and store the analysis result into the database. Finally we convert the hyperlink on the web page to ‘dynamic hyperlink’ during the design phase, which is load from the mining result of the database. This way, we can automatically adjust the web page hyperlink.

Location-Dependent Query is becoming more and more important in mobile environments. In this paper, to improve system performance, author proposes a new model for LDQ. Firstly, a new structure of semantic index is defined, which can store semantic segment circle and timestamp information. Then we analyze the detailed query processing based on our semantic model. To overcome the shortage of FAR at location prediction and influence of semantic valid scope to cache hit ratio, we propose a new improved replacement algorithm (RBF-FAR), in which a cost function P based on semantic segment distance and valid scope, is introduced as replacement criterion. The experiment results show that new model can gain better performance at response time, network-load and cache hit ratio, than the model on the basis of FAR algorithm.

A CPU performance forecasting tool named GcpForecaster is developed for grid environments to help users and schedulers to select computing resources. By running a typical benchmark, GcpForecaster takes periodic measurements of deliverable CPU performance in every grid node and generates performance values. A new forecasting algorithm is adopted to dynamically generate accurate forecasts of CPU performance by using measurement histories. The forecasts can be used to compare different CPUs future performance directly and are made available to users and schedulers at runtime so that they may select computing resources accurately and efficiently.

Page Rank is the basis of the most modern search engines for both Internet and P2P networks, it is defined as the importance of a web page or the reputation value of a peer, viewed as a vote from all the web pages (peers) that link to it, weighted by their importance or reputation. With the exponential growth of the web size and peer nodes of Internet, knowing where to search becomes a really hard problem, optimizing and focusing search is often resolved by computing and comparing page rank values for web page or files. In this work, we present an overview of several top page ranking algorithms for Internet and P2P networks, then analyze and compare their computing performance and precision, finally give some simulation results in power law networks.

Streaming media services like Multimedia services of audio and video are more and more widely applied in wireless networks, and how to maintain this class flows friendly to TCP flows and hold QoS of the streaming service are of great importance. An adaptive rate control mechanism is proposed to solve the problem in this paper. The receiver distinguishes packet loss between network congestion and wireless link error to get the accurate state of the networks. It also combines occupation of the receiver buffer, which reflects how long the temporarily stored streaming medium data can be played. And tradeoff between TCP-friendly and QoS of streaming media service is achieved through multi-level rate adjustment according to packet loss rate of congestion and queue length of receiver. Consequently the high network utility is maintained. The results of mass simulations show that this method is friendly to TCP, stable and of good performance.

Based on the economic models, the current methods of the grid resource management employ the grid user agents to achieve the user resource sharing, the grid resource agents to manage the resources for the owners, the grid resource brokers to coordinate the grid user agents and the grid resource agents. However, these methods constrain the users on the selection and the range of the sharing resources and increase the processing time. This paper proposes another grid resource management model based on GRS (Grid Resource Supermarket) to resolve these problems. A program and a scheduling algorithm for the model are given.

In order to provide a uniform virtual view of resources for different users and applications, a grid computing framework with spaces (GCFS) is proposed. Based on the asynchronous mechanisms of the spaces model, this framework uses spaces to share data for applications. Spaces offer a virtual workspace for both resource providers and consumers in job scheduling. Task assignment is implemented by the coordination between the task manager and task schedulers. The space manager can dynamically create or destroy space services in response to the change of resource supplies and demands. Finally, a simulation of third generation (3G) wireless systems is developed to demonstrate the above ideas, and it is found that a flexible grid computing environment can be realized via GCFS.

Researchers have argued that current Web modelling languages have various limitations including disconnection between business models and technical architectures and also disconnection between functional architecture and information architecture. Addressing the problem; Web Information Exchange Diagrams are developed in two flavours, WIED and WIED-UML. WIED-UML is developed to primarily address the problem of disconnection between functional architecture and information architecture. However, we argue that it, doesn’t really address the issue, because WIED-UML is developed on a set of different UML-Unified Modelling Language notations which don’t have semantic conformity with current UML metamodel. Creating transformation rules between models not conforming to common metamodel vocabulary, would not lead to an elegant solution. Hence in this paper, we propose a solution, using UML 2.0 Profile for WIED. We demonstrate it by an example and argue that it is a potential improvement of generic WIED model besides eliminating the requirement for non-standard WIED-UML.

In order to increase the transmission efficiency of OFDM system, considering the special structure of OFDM system, this paper proposes an EM iteration channel estimation algorithm that uses fewer pilots. The simulation result shows that the algorithm can still converge to the case with given channel parameters when inserting fewer pilots, and the iteration number apparently decreases as the number of pilots increases. The proposed algorithm doesn’t need the statistical characteristic of channels; at the same time, it can realize the trade-off between system performance and complexity by choosing the number of iteration times. The algorithm degenerates to a general EM algorithm when no pilot is used in channel estimation.

A general and integrated method to develop grid workflow applications based on UML is introduced in this paper. Specifying grid workflow vocabularies with UML is illustrated firstly by graphical presentation of grid workflow vocabulary for the ASCI. On the basis of BPEL4WS and GSFL, the problem of specifying Open Service Workflow Language (OSWFL) with UML to design workflow applications in WSRF is addressed. Secondly, it shows that UML can be customized through a ‘UML Profile’ for graphical presentation of grid workflow applications. Lastly, through extension of BPEL4WS mapping, the problem that how to automatically generate a concrete OSWFL application instance from a UML model meeting the profile is explained. In general, both modeling of grid workflow vocabularies and grid workflow applications can be implemented with UML, and the XML-based OSWFL Schema generated from UML vocabulary model can be used to validate the OSWFL application instance, therefore, a special model checking language is avoided.

The resource allocation algorithm has direct impact on the quality of service of grid. Based on the resource allocation framework, this article proposed a QoS resource optimal algorithm for computational grids. By using a two-step optimization strategy, the algorithm can: 1 Accept as more task as possible, allocate the tasks to “best suit” resource, and every dimension resource can be utilized in a balanced manner. 2 Optimize the QoS of tasks; increase the benefit of system resources. The heuristic algorithm fit the current proposed grid resource model, has a better computing complexity, and a simulation experimental has proved its affectivity.

In Mobile IPv6 (MIPv6), triangle routing can be eliminated by means of Route Optimization (RO). BU3WAY protocol is designed to guarantee the secure implementation of route optimization, because it enables the correspondent node to obtain some reasonable assurance that the mobile node is in fact addressable at its claimed care-of address. In this paper we propose a improved method to perfect BU3WAY protocol. The improved method provides a mechanism to make BU3WAY protocol work well when the route between correspondent nodes and home agent or the route between home agent and mobile node is broken without lowering security.

The skyline operator is important for several spatial database applications involving multi-criteria decision making. Skyline computation has recently received considerable attention in the database community, especially for progressive algorithms that can quickly return the skyline points without having to read the entire data file. In this paper we propose an algorithm for skyline queries based on window query, which can prune query space by changing the query window continuously, the number of the accessed data points decreases significantly, and the query result is sound and complete.

An ad hoc network is a kind of wireless communication network which does not rely on a fixed infrastructure and thus lacks in centralized control. Networks of this kind are based on the fundamental assumption that all the nodes would cooperate and not cheat. This assumption becomes invalid in the situation that compromised or malicious nodes exist. In this paper, we propose a Bayesian network-based trust model and a method for building trust based on the recommendations from other nodes and the node’s own experiences. The model is suitable for ad hoc network without a dedicated trust infrastructure or pre-configuration of nodes. Trusted route can be easily found out and malicious nodes can be circumvented by using the trust model.

In this paper, a FNN-based adaptive control system for a miniature climbing robot is presented. The main components of the system include a multi-sensor data fusion model and a FNN-based adaptive controller, which consists of a five-layer FNN, a fuzzy-rule base and an adaptive gait sequence generator. By this control system, the robot can autonomously generate proper gait and control itself to transit between two differently inclined surfaces. The properties of the FNN-based adaptive control system are verified by the computer simulations.

The Octopus card is the most popular smartcard in HKSAR, China. This article analyzes the business development of October card from core-transport payment, to other transport payment and then towards retail payment and beyond. It provides lessons learned from a IT-based business transformation perspective and evaluates the key success factors of its the rapid market penetration and sustainable developments within and outside HKSAR.

Machining complex freeform surfaces more efficient and more flexible has been a challenging task. The equivalent normal curvature milling model is a new milling method proposed in this paper based on 5-axis CNC bed-type horizontal milling machine. At the same time, a specialized whirlwind milling tool is also introduced. The tool path planning of this new milling method emphasizes the alignment of milling tool orientation so as to approach to the surface being machined in the form of equivalent normal curvature envelopment. So this new method will minimize the difference of normal principal curvatures between the surface being machined and the symmetric rotational surface of the milling tool and improve the metal removal rate (MMR). The new whirlwind-milling tool with circular standard-sized cutters is easy to enhance the cutting finish and productivity. The cutter path description and machine-tool coordinate transformation are analyzed, and some examples about this new equivalent normal curvature milling method are discussed.

Autonomous planning system for spacecraft is a crucial issue for many space applications. Concurrences for actions and complexity of constraints for autonomous planning problem of spacecraft have been analyzed, and the main elements of planning model have been analyzed. Subsequently the modeling approach for planning domain is given based on Planning Domain Definition Language, the representation for duration constraint has been proposed based on interval theory, and timeline models have been adopted to represent the movements of spacecraft. Finally, a simplified example is given to demonstrate the modeling of autonomous planning system for spacecraft and representation of corresponding constraint.

Thin client system is very important to those network based applications, such as network computer system. This paper proposed a new conception, called service net, to organize and manage multi-server based thin client systems. For multi-server extension, general method and service net based method are compared. With service net, multi-server based thin client systems can be well managed by the policies for user account management, redundant storage, request distribution, load balancing, fault tolerance, fault shifting, dynamic joining and quitting. A prototype system based on Tsinghua Network Computer System is implemented. The results show that QoS of service net based method is more stable and better than the one of general method.

Sensor management is one of the most important parts of multi-sensor data fusion system. In this paper, we present a sensor management method based on total discrimination gain in multi-sensor situation assessment system. The total discrimination gain includes probabilistic discrimination gain and fuzzy discrimination gain. We generalize the probabilistic discrimination to total discrimination. This gain is used to select more informative sensor. This approach is proved to be effective by an instance.

This paper has analyzed the situation of existent fire alarm systems and has proposed a new method of connecting local fire alarm controllers to build a much larger and more powerful system, with GSM (Global System for Mobile Communications) as the communication medium. All these local fire alarm controllers, which work as slave stations, and a central monitor, which works as a master station, are incorporated in the GSM network. Information is transmitted between the master station and slave stations through SMS (Short Message Service). The general structure of the system is presented in this paper. It is a three-hierarchy model and it consists of three subsystems: detecting and controlling subsystem, communication subsystem and central monitor subsystem. The function of each subsystem is explained and the hardware and software details of the communication subsystem are described.

The routing management styles of Information System Project (ISP) were surveyed, and the two basic solutions of process control were discussed. The concept of Process Pattern is suggested based on the analysis of strategies of process management. Some typical process patterns are analyzed and the XP Pattern is summarized.

Control and scheduling theory is integrated to enable the complex control algorithms with varying execution times in embedded real-time control systems (ERTCS). A case study of the ERTCS that consists of a set of model predictive controllers is given in this paper. A feedback scheduling approach is presented to optimize the global performance of ERTCS subject to limited computational resource. The scheduling algorithm is shown robust against the varying of execution time of MPC tasks at runtime. Simulation results illustrate its effectiveness.

This article is aimed at researching the application of IPv6 technology to the RCM2250 embedded system, which is applied to the communication technology in the Ethernet. Based on analyzing and studying the inlaying TCP/IP stack of MPU, the Open Source codes of Dynamic C and the standard of the next generation protocol IPv6, IPv6 stack has been realized by modifying and adding the library files of the programming language Dynamic C in the RCM2250 embedded system. The stack includes such module as data receiving and sending, neighbor discovery, ICMPv6, etc. mainly, and the stack’s APIs make the embedded equipments communicate with the industrial computer through the Ethernet easily.

Currently, the false positive and the false negative of Intrusion Detection System are very high. It was always the main problem that bothered the user of IDS. The two main technologies-Anomaly Detection and Misuse Detection are applied in IDS. Because both the technologies have its own advantages, IDS combined with the tow technologies was used more and more widely. To this problem, this paper presented a model of IDS based on combination of Misuse Detection and Anomaly Detection. In this model, Misuse Detection is based on pattern matching and Anomaly Detection is based on statistical analysis. It combined the tow technologies to reduce the false positive rate and the false negative rate, and then to improve security of IDS.

More and more complex embedded system such as network processor use component based development. Non-functional properties of these embedded component models are used to predict performance of the final system. Component in network processor based system has some special features, which are pertinent to the hardware architecture of network processor and computational model of applications. We proposed a non-functional property, i.e. thread property of the component model, and on which we developed a performance analyzing method. This method provides quantified component performance, and then the worst-case performance of final system can be predicted at composing time. Case study of IPv4 component shows that there is a reasonable difference between theoretic analysis and simulation result. The proposed method is based on network processor, but the basic concept could also be applicable for other similar systems.

As a result of development of embedded system, the concept of embedded system is improved from “the computer system embedded in large system” to “the system completing independent functions.” At first, the many application instances of embedded systems are listed, including the fields of aviation, aerospace, military affairs, vehicle embedded, communication, the main features of application environment of embedded system are given. Then, a survey of popular embedded operating systems is presented. And then the future trends of embedded system are analyzed from three aspects: open & standard & COTS; real-time & reliability & security; computing capability & application environment. Finally, the main research topics of embedded system are given.

In this paper a design thought that based on multi-thread, data buffer, communication techniques is proposed by which to improve real-time capability of communication for configuration software, a multi-thread model is designed to solve the problem of time-consuming I/O operation; The communication instructions in systematic will divide into two pieces of PRI, and cooperate with the data buffer queue of first-in first-out with PRI. It ensures that writing instruction can be executed first. An approach that combines homogeneous communication instruction and improves real-time capability of communication is presented in this paper. Finally, the results which are achieved with improved communication module show that the approach is valid on the running-platform.

Present applications of artificial intelligence technology for wastewater treatment in china are summarized. Expert system was mainly used as the system operation decision-making and fault diagnosis. Artificial neuron network was used as system modeling, water quality forecast and soft measure. The application of fuzzy control technology was the main aspect of intelligence control in wastewater treatment process. Finally, the main application problems of artificial intelligence technology for wastewater treatment in china are analysed.

This paper presents a Windows based system for monitoring wood piece materials on a conveyor belt. To make the system work efficiently in plants, the system was designed and developed with a flexible Windows platform which mainly consists of image acquisition, image processing, object delineation and analysis, and interface between different computers in use. A number of functions for image processing and analysis are included in the system, A number of newly-developed algorithms can delincate wood pieces with high accuracy and high speed, and in the wood piece analysis part, each wood piece can be characterized by 50-60 parameters that can also be used for constructing wood piece models directly in a Windows platform environment. The system has been tested and used.

A novel reduced support vector machine on Morlet wavelet kernel function (MWRSVM-DC) is proposed, which combined the Morlet wavelet kernel function with the reduced support vector machine (SVM) model properly. Based on the wavelet decomposition and conditions of the support vector kernel function, Morlet wavelet kernel function for support vector machine (SVM) is proposed. At the same time, because SVM is restricted to work well on the small sample sets, a novel reduced SVM on density clustering (RSVM-DC) is proposed. This algorithm focuses on dealing with a sample set through density clustering prior to classifying the samples. After clustering the positive samples and negative samples, the algorithm picks out such samples that locate on the edge of clusters as reduced samples. These reduced samples are treated as the new training sample set used in SVM’s classifier system. Experiment results show that not only the precision but also the efficiency of SVM’s are improved by MWRSVM-DC.

More and more experimental results have shown that genes do not function independently; instead they act on each other. To find the interactions among genes is one of the hottest topics in current genome research. The Bayesian Network (BN), which is a graph-based representation of a joint probability distribution that captures properties of conditional independence between variables, is a desirable tool to find the interaction between genes. However, how to find appropriate BNs that most fit to the data is very difficult since the number of possible BNs on n variables is the super-exponential of n. To avert the combinational explosion, in this paper, we use Estimation of Distribution Algorithm (EDA) to search the BN space. Also, in order to make the individuals of EDA meaningful, we also propose a depth-first search method to cut circles in the graph. We have tested our method on cell-cycle gene expression data, the results show that the constructed BNs can not only discover some existing relationships in other literatures and Gene Ontology, but also reveal some previously unknown interactions.

The processing of Chinese entity names is significant to automatic text processing. It is a more challenging issue in Chinese text processing than English text processing. The paper presents a hybrid approach which automatically recognizes named entity from Chinese text. This approach combines the heuristics, proper noun lexicons and statistical information drawn from a Chinese corpus. The experiment based on Internet news shows that the F-measure reaches 89.85%.

As an empirical method, EBMT avoids complicated deep-level linguistic and semantic analyses, provides a new way for machine translation. The use of bilingual aligned trees has improved the adaptability and accuracy of EBMT. In this paper we introduce concepts and developments of some important EBMT models based on the bilingual aligned trees.

The automatic extracting of key sentences is one of fundamental works of automatic text processing, such as text compressing, automatic abstract et al. The paper presents the approach and the processing procedures for extracting key sentences from text. By our method, sentences are rank by a weighted combination of statistical and linguistic metrics. We put the emphasis on how to form the weighting schemes and acquiring the experiencing coefficient. For minimize redundant sentences, we outline a new algorithm choosing sentences from candidates. The experimental result shows that the ratio of the sentences extracted as very relevant, somewhat relevant and non relevant to the text topic is 63.05%, 28.15% and 8.8% respectively. For evaluating the kind of application further, we propose a novel more objective evaluation method based on the similarity and the diversity between manual extraction and automatic extraction.

An improved particle swarm optimization (IPSO) algorithm is proposed to solve a typical combinatorial optimization problem; traveling salesman problem, which is a well-known NP-complete problem. In the improved algorithm, particles not only adjust its own flying speed according to itself and the best individual of the swarm but also learn from other individuals according to certain probability. This kind of study behavior accords with the biological natural law even more, and furthermore helps to find the global optimum solution. At the same time, this paper proposes the concepts of Adjustment Operator and Adjustment Sequence based on which particle swarm optimization (PSO) and IPSO algorithm were successfully rebuilt, according to the ideas of single node regulating algorithm. For solving traveling salesman problem, numerical simulation results show the effectiveness and efficiency of the proposed method.

This paper proposes the architecture of an agent-based hybrid recommender system consists of three subsystems: content-based, collaborate filtering and rating prediction. They are controlled by a task agent, who self-determines when to execute the hybrid recommendation algorithm based on user’s implicit feedback and its knowledge base. In content-based subsystem, an improved implicit feedback algorithm is presented. Moreover, an improved inter-user similarity function is applied to find out like-minded neighbors in collaborate filtering subsystem. To enhance the quality, result sets of content-based and collaborate filtering recommendations are reunited to select the top 5 items through our rating prediction algorithm based on linear regression model. Afterwards, a client agent sends back recommendations and receives another user’s request. Experimental data show that hybrid techniques hold the promise of produce high-quality recommendations with the assistance of agents.

This paper studies the problem of data association in multiple maneuvering targets locating and tracking. Genetic algorithm is used successfully in solving the complex optimization and the industrial engineering problem. Recently researching on genetic algorithm has attracted a lot of attention. This paper puts forward a way of using Genetic Algorithms to resolve the problem of multiple target data association. The simulation results show that the algorithm used in this paper is able to avoid the local extremum and the outcome is satisfactory.

The mission of Earth Observing Satellite (EOS) is to meet the imaging requests of scientific observing, mapping, military reconnaissance and commercial activities. Satellites scheduling is an important real world problem that impacts the use of expensive and limited resources; it is belongs to a class of multi-machines scheduling problem with time window constraint. The objective of satellites scheduling problem is to make the maximum efficiency of use of satellites. It is proved to be NP-hard in computational complexity. Many traditional techniques are tried to solve this scheduling problem but fail in the large search space and not well understood bottleneck. In this paper, the constraints of satellites scheduling problem are presented, and an overview of the studies of evolutionary computation in satellites scheduling domain also be represented. This paper pointed out that evolutionary algorithm could yield good performance in solving EOS scheduling problem, especially in much more complex condition. That implicates that our future research could adopt evolutionary algorithm.

A new Term Co-occurrence Model based on Vector Space Model (VSM) for improving the performance of text classification is proposed in this paper. The model selects a whole document as basic unit for obtaining term co-occurrence resources, which are then applied to text classification. Experimental results show that it improves the performance of text classification system greatly.

In machine learning, inductive techniques perform a global approximation to the target concept. Instead, lazy inductive techniques use local approximations to form an implicit global approximation of the target concept. LID (Lazy Inductive Description) and C-LID are lazy inductive techniques in feature term structure representing case-based reasoning, and behaves very well in the Predictive Toxicology Challenge (PTC). The main contribution of this paper is the proposal of three policies in feature term structure representing case-based learning, which underlies an improved technique — SSU-LID. The outcome of SSU-LID is assessed with experiments on the Toxicology dataset.

Lazy learning algorithms do not construct an explicit hypothesis, and inductive process can be attributed to the classifier which must use the training set to map each instance to its label. Many researches were done about lazy learning, most of which is focus on the classic classification. In this article, we apply lazy learning algorithm to ordinal classification which exemplified by ordinal decision tree. By taking one sample example, we describe the procedure of lazy ordinal decision trees induction and give the theoretical analysis.

A process is described and evaluated for segmenting and clustering continuous audio streams automatically by speaker with no prior speaker model. The process consists of two steps: (1) segmentation based on GLR distance measure and BIC refinement, (2) clustering based on agglomerative clustering and pruning selection. The aim is to produce just one pure cluster for every speaker. Results are presented using utterances derived from the 2002 NIST development data.

An adaptive step length genetic algorithm (ASLGA) is presented, in this paper, to improve local search efficiency and to prevent premature convergence of the genetic algorithms (GAs). Step length of crossover and mutation are adaptively varied depending on the fitness values of the solutions to improve local search efficiency. Similar solutions are adaptively disrupted by multipoint mutation according to the livingness index of the population to prevent premature convergence. Three typical multimodal functions are used to verify the performance of the ASLGA. Experimental results demonstrate that the ASLGA has good local search efficiency and global search capacity.

Fingerprint identification is one of the most reliable personal identification methods and it plays a very important role in forensic and civilian applications. In this paper, a novel method of fingerprint identification based on the features extracted from the integrated Discrete Wavelet and the Fourier-Mellin Transform (DWFMT) is proposed. Discrete Wavelet transform is used to smooth and preserve the local edges after image decomposition, and hence making the fingerprint images less sensitive to shape distortion whilst Fourier-Mellin transform served to produce a translation, rotation and scale invariant feature. Multiple DWFMT features can be used to form a reference invariant feature through the linearity property of Fourier-Mellin Transform and hence reduce the variability of the input fingerprint images. The experiments show the identification accuracy is over 96% and 99% of identification rate is achieved when multiple DWFMT features are used.

Bayesian networks are powerful tools for probabilistic representation and inference, their using as classifiers has received considerable attentions. However, much work in past focused on Bayesian scoring techniques for learning BN, which leads to poor performance due to the mismatching between the objective function and the goal of classification (maximize accuracy). In this paper, we propose an Error-driven stochastic learning algorithm which search for more accurate classifier based on classification error. To improve efficiency and robustness, an improved Markov Blanket discovering algorithm based on incomplete dataset was applied to feature selection in advance. Experiments were conducted on several datasets and the results showed that this approach can achieve better classification accuracy even with missing data.

Three linear and non-linear dimension reduction algorithms, LSI, Isomap and SIE for processing high dimensional data in natural language documents are presented in this paper. Document data processed by the three algorithms is then clustered by bisecting K-means clustering algorithm. Experimental results show that the performance of SIE algorithm and LSI algorithm in document clustering are basically equivalent and better than the benchmark, while the performance of Isomap algorithm is worse than the benchmark.

The paper presents an approach for Chinese text summarization. Unlike normal statistical methods, we capture concepts of words based on HowNet, and construct conceptual vector space model by using concept (word sense) instead of words as the feature. According to the conceptual vector space model and pos-information, each sentence of a document is rated. Sentences with high weights are selected as rough summarization, and then degrees of semantic similarity of sentence are calculated for reducing the redundancy of summarization. In order to evaluate the summarization system, we use two different methods, intrinsic and extrinsic approaches. The results show that these approaches are effective, and performance of the system is promising.

Automating schema matching is challenging. Previous approaches to automating schema matching focus on computing direct element matches (1:1 matches) between two schemas. A schema matching system structure CMatch is constructed in this paper, which discovers both 1:1 and complex matches. It does not discover matches between schemas roundly, but also improves the efficiency and accuracy compared with other schema matching systems in practice. Compared with iMAP, about 60 percents efficiency and 11 percents accuracy are improved in practice.

In this paper, we propose a new and practical definition of language---language is a kind of map. This definition reflects both the descriptive and explanatory properties of the human language under the framework Chomsky proposed. It becomes easy to clearly describe the process of language generation and understanding using this definition of language. And furthermore, because this definition makes the natural language and the artificial language under one umbrella, it becomes clear why and how the fundamental difference between the natural language and the artificial language---ambiguity exists. All these mentioned can be explained by our new definition of language. This definition formalizes most of the aspects of the natural language, so it not only has the explanatory significance in understanding language, but also has the practical significance in better understanding and designing the natural language processing systems. It is believed that only when a unified and clear definition of what our research object---language is in NLP field, can significant advances be made.

In this paper a new method is presented to solve a series of fault detection problems using Probabilistic Model (PM) in Genetic Programming (GP). Fault detection can be seen as a problem of multi-class classification. GP methods used to solve problems have a great advantage in their power to represent solutions to complex problems, and this advantage remains true in the domain of fault detection. Moreover, diagnosis accuracy can be improved by using PM. In the end of this paper, we use this method to solve the fault detection of electro-mechanical device. The results show that the method uses GP with PM to solve fault detection of electro-mechanical device outperforms the artificial neural network (ANN).

In the automatic text classification, seeking an efficient feature selection algorithm is still one of the most important problems and needs prompt solutions. When selecting feature selection algorithms, both the algorithm complexity and the feature selecting effect should be taken into comprehensive consideration. In this paper, a novel feature selection algorithm, which is based on TFIDF-weight and KL-divergence, is designed and implemented after considering the TFIDF weights of feature words. In comparative experiment, Mutual Information method is taken as evaluation criteria. The experimental result is quite satisfying, for the features selected by this algorithm can not only better stand for the categories but also diminish dimensions of feature space effectively.

We present an approach based on lexical examples and semantic collocation for resolving Chinese Formalized Classifier selection. We discuss the characteristics and collocation abilities of Formalized Classifiers. The semantic representations of nouns collocated with Formalized Classifier in the approach are based on the lexical concept definition in HowNet. We will also show the application of the approach in our machine translation system based on semantic language. Our test results indicate that most of the Formalized Classifiers can be selected correctly in target language generating by this approach.

An adaptive multi-agent ACO clustering algorithm is proposed in this paper to enhance the efficiency and quality of ant based clustering. The algorithm includes three levels: Level-0 agents build solutions, level-1 agents improve solutions and level-2 agents update pheromone matrix. The updated pheromone then provides feedback information for the next iteration of solution construction. Mutation probability and pheromone resistance are the adaptive parameters, which can be adjusted automatically during the evolution progress to solve the contradiction between convergence speed and precocity and stagnation. Experimental results show that the proposed algorithm is more effective compared with the clustering algorithm based on GA and k-means, and the clustering quality and efficiency are promising.

The main objective of the paper is to introduce a new algorrithm of attribute reducts based on variable precision rough set theory. An improved genetic algorithm (GA) which adopts information entropy as it’s fitness function is introduced. The strategy of mixed crossover and two points mutation enlarges the search scope. The cross generation elicit selection and self-adapting strategy make the genetic algorithm converge to the overall optimal solution stably and quickly, which gives it an edge over the normal GA. The effectiveness and the advantage with respect to the norm GA are checked though an example.

Word similarity computing plays an important role in the natural language processing (NLP) field. HowNet system as a semantic taxonomy (or ontology) is a very good system in the NLP field. This paper first introduces the method of word similarity computing based on HowNet. From the introduction we draw a conclusion that the speed of word similarity computing depends on the sememes distance computing. Then it presents a method of code sememes which aims to make the sememes distance computing easy and quick. After this it proves the correctness of the algorithmic and gives the algorithmic of sememes distance computing which is based on sememes coding. At last it discusses the benefits and disadvantages of the algorithmic.

Cerebellar Model Articulation Controller (CMAC) is a kind of adaptive ANN based on local learning, which has the characters of fast convergence speed, strong generation ability, simple structure. While its advantages over conventional control techniques have been recognized in the literature, it has primarily been in system identification and pattern recognition, but not in control application. A novel adaptive learning algorithm and concurrent control based on CMAC and PID is proposed in this paper. The proposed algorithm and controller are applied to two-order nonlinear system. The experimental and simulated results show that performance of the CMAC based controller using the proposed learning algorithm is more stable and effective than that of the conventional controllers.

A novel Radial Basis Function (RBF) network was proposed, which combined ant colony clustering (ACC) and radial basis function with artificial neural network (ANN) model properly. The approach utilized ant colony clustering to find the each kernel functions’ center of RBF and use a new method to reduce the neuron’s number of the hidden layer. It avoided potential serious local optimal result caused by local trap and could achieve a more accurate RBF network. Additionally, the efficiency of RBF network’s training procedure was improved. It constructed a universal and efficient framework of RBF network without depending on the original selected centers of each cluster. Experiment results show that the precision is increased, the speed of training RBF network with this new algorithm is also accelerated greatly, and the structure of this novel RBF network is simpler than other algorithms.

The automatic editing with the improvement of content and efficiency becomes urgent, especially for the dictionaries and thesauruses, as they possess regular formats and immense size. However, Chinese characters and words do not have definite alphabetic orders as English words are, it’s always difficult and time consuming to establish the index in the process of compilation. We propose an approach to establish automatically index of entries through Chinese phonetic alphabets and radicals, and then present the new model of index system based on the definition of basic index. Our method can facilitate structuring an automatic editing system and greatly raise the efficiency of compilation.

A reconfigurable cipher chip has higher security and flexibility than a conventional one. Therefore, it will be an important part of the future security mechanisms. For designing such kind of chips, the challenge is to improve the throughput rate. In this paper, the reconfiguration of DES, 3DES and AES is analyzed firstly. Then, it proposes a new reconfigurable architecture based on pipelining and paralleling which can implement the three algorithms with a high-speed. The results show that the operating frequency is 100MHz and the throughput rate is 7Gbps for DES, 2.3Gbps for 3DES and 1.4Gbps for AES. Compared with the similar existing implementations, our design can achieve a higher performance.

Electronic data interchange made significant contributions to productivity and inventory control in electronic business (e-Business). The XML-based e-Business specifications are the main stream in B2B (Business to Business) field. Hundreds of industry-defined XML (Extensible Markup Language) schemas were created over the last year, and the ebXML (electronic business XML) specification is the most promising one among them. This paper addresses data interchange from EDI (Electronic Data Interchange) to ebXML in business processes. It presents how e-Business technologies, such as ebXML, enable enterprises to collaborate at the business process level. It allows organizations to conduct businesses electronically using the Internet. The development trend and the recent implementations are shown by two Chinese official projects on the basis of ebXML as examples.

It is very difficult to discover valid knowledge in the data repositories and is also very difficult to choose suited data mining methods without prior knowledge about data mining or application domain since the amount of raw data becomes large and there are a variety of data mining methods. In this paper, we propose a framework of an Intelligent Knowledge Discovery System (IKDS) to help users select appropriate data mining algorithms and discover knowledge. In addition, a knowledge acquisition methodology, SIMCAP, is also proposed to elicit not only explicit knowledge but also implicit knowledge of the experts. The knowledge in IKDS can be represented and stored by XML. The prototypes of SIMCAP and IKDS have been built up to help users discover knowledge.

A clustering algorithm based on swarm intelligence is systematically proposed for anomaly intrusion detection. The basic idea of our approach is to produce the cluster by swarm intelligence-based clustering. Instead of using the linear segmentation function of the CSI model, here we propose to use a nonlinear probability conversion function and can help to solve linearly inseparable problems. With the classified data instances, anomaly data clusters can be easily identified by normal cluster ratio. And then the identified cluster can be used in real data detection. In the traditional clustering-based intrusion detection algorithms, clustering using a simple distance-based metric and detection based on the centers of clusters, which generally degrade detection accuracy and efficiency. Our approach can settle these problems effectively. The experiment result shows that our approach can detect unknown intrusions efficiently in the real network connections.

To predict SARS epidemic spread with more precision, the adaptive neural network with the adaptive filter is used to predict SARS epidemic in this paper. With the adaptive BP neural network, we set up the relationship between the previous daily suspect, probable cases and the probable cases of next day, with the adaptive filter, we have got the more precision of the prediction. Finally we have predicted the probable cases in Beijing from May 22, 2003 to June 10, 2003, the result is satisfied.

The research of modeling agents is a basic work of agent system and multi-agent system. So far, a large amount of models have been proposed, such as the models based on the mental states (BDI) of Agents, the models based on the function, which maps environment states or percepts to action, and the models based on the game theory. However, almost all of the methods merely model agents from one aspect. Thus, they don’t meet the need of complex social actions of agents. (i.e. coordination, cooperation, negotiation, etc.) Therefore, in this paper, we combine the external environment state with agents’ mental states and utility. And we propose an agent model, in which not only the history of environment and the history of action are considered, but also the mental states and decision utility of agents are considered. The balance between rational consequences and rational decision utility of agents was found out. We improve the models proposed by Russell, Wooldridge, Rao, Xu etc to make the model of agents meet the need of agents’ interaction with others in the complex social environments.

Unsupervised learning strategies such as self-organizing map (SOM) may be more fascinating in some applications in which lacks of supervised signals. But the standard SOM network has a fixed number of outputs that must be pre-specified before training, which lacks of flexibility. In this paper, a fast clonal selection algorithm (FCSA) for constructing an immune neural network (INN) is presented based on clonal selection principle. The INN is a two-layer network whose number of outputs is adaptable according to the task and the affinity threshold. The constructed INN is similar to the ABNET proposed by de Castro et al, but the constructing algorithm FCSA is remarkably simpler, faster, and more facile than that of ABNET, which can be demonstrated in the simulation experiments.

In this paper, we present an application of rough set methods for rule generation, which includes two steps. Attribute reduction is done first, which reduces the number of unnecessary attributes for the whole decision table. In this process, the relative subdivision distance is proposed as the heuristic information of algorithm REDA. The completeness and the correctness of REDA are also proven from the theoretical analysis. Value reduction is performed afterwards, which reduces the number of redundant attributes for every object. In this process, properties of attribute values of three classes are analyzed and the maximum number of condition attribute values is removed without losing essential information according to their properties. We then can get the optimal decision rules. Experimental results from applying the rough set approach to the set of data samples are given and evaluated to show efficient and effective of our algorithm.

In the purpose of using the well-formed results in rectangle algebra to solve the problems in cardinal directions, this paper discusses the correlations between rectangle relations and cardinal directions, then gives two algorithms to check the consistency and pairwise compatibility of the cardinal directions. Through the proof of correctness, the author points out that the path-consistent is sufficient to judge the consistency of basic cardinal direction constraints to the regions, which are connected and have connected boundaries. Finally, the examples are presented to verify the algorithms.

The increasing number of knowledge-based systems that build on Bayesian networks and dynamic Bayesian networks acknowledges the usefulness of these frameworks for addressing complex real-life problems. The usually large number of probabilities required for their application, however, is often considered a major obstacle. The use of qualitative abstractions may to some extent remove this obstacle. Qualitative Bayesian networks and associated algorithms have been developed before. Based on qualitative Bayesian networks, in this paper, we present three definitions for the time-series dynamic Bayesian networks with feedback loops and qualitative time-series dynamic Bayesian networks by defining qualitative influence between adjacent time slices analogously. Then, we apply the qualitative dynamic Bayesian networks with feedback loops to an economic example and make qualitative decision successfully.

Based on the GUO’s Algorithm, a high-efficiently hybrid evolutionary algorithm is proposed. The new algorithm has two main characteristics: first, introduce the Gauss mutation operator of Evolution Strategies (ES); second, introduce variable searching subspace. In order to avoid premature of population, the Gauss mutation operator is used; at the same time, for accelerating convergence, the searching subspace can be reduced automatically when the population’s evolutionary value is very close to the global best value of the population. Numerical experiments show that the new algorithm is high-efficiency and the precision of results is very high, at the same time, the experiments’ results of several test functions exceed the best value recoded in the references.

The purpose of this paper is to search for a prediction method of the gas outburst and gas emission quantity of coalmine. Extracting sensibility parameters and using sorting scheme of the artificial neural network (ANN), the paper solved forecasting question of coal & gas outburst which mathematic model is built up difficult. Research shows that ANN is a good method for forecasting coal & gas outburst and gas emission quantity, which sets up foundation for safety and production decision support system (DSS) of coalmine.

The technique of relevance feedback is one of the most popular methods of query optimization in information retrieval (IR). Among the existing methods of query optimization, some use directly the feedback to form their optimized queries, while others use genetic algorithms (GAs) to optimize their queries according to the feedback. Combined with a GA and the feedback, a new method of forming optimized query has been presented in this paper. Comparison experiments of our method with the four well-known methods (Horng & Yeh’s GA method, López-Pujalte et al.’s two GA methods and Ide’s traditional Dec-hi method) have been given on the four benchmarks (Cranfield, Medline, CISI and NPL collections). The test results show that our method can achieve better results than the other four.

Anti-attack ability is one of the key restricted factors for applications of digital video watermarks. In the practical environment, the geometric and temporal distortions are commonly used to invalidate the video watermarking intentionally or unintentionally. As a countermeasure, a novel video watermarking that can resist the spatio-temporal desynchronization simultaneously is presented in this paper. The feature of the transform domain that will not be altered easily with the geometric distortions is investigated. Then the watermark information is embedded via altering this feature along the temporal axis. In order to preserve the temporal synchronization, the sequences of spread-spectrum that have good correlation properties are employed. Experimental results show that the novel video watermarking proposed can tolerate the large geometric distortions. The synchronous resistance to the temporal distortion is also proved in the experiments.

In EDSS there have not only temporal data but also temporal knowledge. Temporal repression is a very important problem. This paper describes a simple and maneuverable model framework. It uses the interval algebra theory to reconstruct knowledge base, adopts the layered conception of case character factors to construct cases, and integrates temporal reasoning and cased-based reasoning. This framework can realize knowledge’s temporal properties and corresponding relation between the set of character factors and the set of rules. It implements temporal knowledge expression and reasoning and provides dependable proofs for decision-maker. The paper final target is trying to solve the knowledge temporal problem in dynamic decision-making environment and provides an applied platform for researchers.

Input variable selection is always a significant problem in nonlinear system modeling. In this paper, we propose an effective and efficient input variable selection method based on Taylor series. We assume the nonlinear system as an at least 3 order derivative and continuous function f(x), and approximate it with 3 order Taylor series in the small proximities of the sample data. We define the average absolute values of first order partial derivates at the sample data as the importance indexes for the input variable candidates, and the average absolute values of second order partial derivatives as the correlation degrees between the input candidates. The importance indexes of the input variable candidates can be calculated by solving the linear equation at low computational cost. In this paper, we describe the nonlinear system as an Adaptive Network Based Fuzzy Inference System (ANFIS). Guided by the priori knowledge of input variables, importance indexes and correlation degrees, a fast input variable selection method is presented in which Regularity Criterion is adopted to evaluate the performance of the selected input variables. Experiment results and theoretical analysis both prove that this input variable selection method can speed up the selection process and is more powerful than the typical search tree method.

According to the characteristics of acoustic fault sources of underwater vehicles, a novel sources detection model using radial basis function (RBF) neural network based on fuzzy clustering algorithm is proposed. The extended fuzzy c-means (FCM) clustering algorithm is utilized to determine the number of hidden neurons, especially, the output layer neurons can be modified on-line so that the network has the capability of incremental learning. An example of diagnosis indicates that the proposed neural network diagnosis system can detect and recognize new faults, and can learn incrementally.

This paper introduces a credit risk evaluation system of financial organization based on CBR and RBR, which combines the advantages of the two reasoning approaches. It focuses on the realization and improvement of CBR in this system, and introduces the algorithm of case matching similarity and the improvement of the case base maintenance approach. Meanwhile, it points out some questions of this system and brings forward the further study directions.

An implement methodology with ontology for Knowledge Element based knowledge storing and search is proposed. Ground on the formalization theory of ontology, design the ontology, relation, axiom, studied two Knowledge Element search approach and feasible technique for knowledge system implementation. This approach can provide a knowledge system with more powerful abilities on semantic expressivity, intellective search and more supporting on knowledge sharing, reuse, interoperability.

One of the major obstacles in querying distributed textile data involves semantic heterogeneity. Our work contributes to the Semantic Web and to textile information interoperability. This paper presents a sort of textile information sharing system architecture and ontology solution by the example of three-dimensional body scanning data sharing and apparel classify retrieve. It describes our system that provides textile information sharing, publishing and retrieving over heterogeneous textile data sets distributed over the Web. We focus on resolving semantic heterogeneity at the value level to accommodate distributed data sets having conceptually similar attributes in which values are drawn from diverse domains. The proposed system can effective overcome the structure and semantic heterogeneity of local three-dimensional body scanning database. By using ontology technology and domain rules, it can also offer intelligent and semantic retrieve tools.

The resource leveling problem in Network planning is a combinatorial NP problem. In this paper, in order to give the neural network description of resource leveling problem, the new concept of Augmented Permute Matrix (AMP) is proposed. Some novel technologies are using when setting up the energy function under time and resources constrains. An Embedded Hybrid Model combining Discrete-time Hopfield model and SA (DHNN-SA) is put forward to improve the optimization in essence in which Hopfield servers as State Generator for the SA. The comparison with professional project management software shows that the energy function and hybrid model given in this paper is highly efficient in solving resource leveling problem to some extent.

Radial Basis Function Neural Networks (RBF NN) are frequently used for learning the rule of complex phenomenon and system. But kernel matrix computation for high dimensional data source demands heavy computing power. To shorten the computing time, the paper designs a parallel algorithm to compute the kernel function matrix of RBF NN and applies it to the prediction of converter re-vanadium modeling. This paper studies the possibility of using parallel kernel RBF regression for modeling an intelligent decision system for re-vanadium in metallurgical process. The paper then implements the algorithm on a cluster of computing workstations using MPI. Finally, we experiment with the practical data to study the speedups and accuracy of the algorithm. The proposed algorithm proves to be effective and practicable in its application.

There are growing interests in algorithms over data streams recently. This paper introduces the problem of sampling from sliding windows of recent data items from data streams and presents two random sampling algorithms for this problem. The first algorithm is a basic window-based sampling algorithm (BWRS Algorithm) for count-based sliding window. BWRS algorithm extends classic reservoir sampling to deal with the expiration of data elements from count-based sliding window, and can avoid drawbacks of classic reservoir sampling. The second algorithm is a stratified multistage sampling algorithm for time-based sliding window (SMS Algorithm). The SMS algorithm takes different sampling fraction in different strata from time-based sliding window, and works even when the number of data items in the sliding window varies dynamically over time. The theoretic analysis and experiments show that the algorithms are effective and efficient for continuous data streams processing.

Purpose RBF ANNs are being applied more and more frequently, especially in pattern recognition. But it is quite difficult to determine the number of hidden-layer nodes and every hidden-layer node’s function. Based on a fuzzy automatic clustering algorithm, a construction method of pattern recognition RBF ANNs is presented in this paper. Technology and Methods First, the given swatches are clustered with the fuzzy automatic clustering algorithm. How many clusters there are, how many hidden-layer nodes are needed. Every cluster’ center is its correspondent radial basis function’s center. Every radial basis function’s width isn’t less than its correspondent cluster’ radius. Second, because only 1 swatch is recognized every time, there is only 1 input-layer node. Third, how many patterns the given swatches may have, how many output-layer nodes are needed. If all given swatches can only be normal or abnormal, 2 output-layer nodes are enough. Results and Conclusions The construction method is introduced in detail with an example, and a RBF ANN is constructed. Good pattern recognition RBF ANNs can be constructed step by step with the construction method presented in this paper.

To overcome the fault of the existing algorithms used to solve the 0-1 assignment problem, this paper presents a new algorithm—Gradually Optimizing Adjustment Algorithm. Through adjusting the Assignment Matrix gradually, the algorithm aims to make the time, which is spent by the worker who spends the most time, shorter by degree. When adjusting, we adjust firstly the job whose interval of the time workers need is the largest, and finally achieve the optimal effect. Through experiment with stimulated data and comparing the result showed in figure in statistics method, we test and verify our algorithm.

To meet the requirements of knowledge classification and process of knowledge updating or modifying frequently in a reconnaissance vehicle agent, a formal bottom-up Object Oriented Fuzzy Petri Net (OOFPN) modeling method is used in this paper. It not only cares about what reasoning rules are being processed in a reconnaissance vehicle agent, but also pays much attention on the structure, subsystems and how they communicate each other. Based on the analysis of a reconnaissance vehicle agent model and its knowledge reasoning system, the modeling theory, and a case of OOFPN are studied. The example demonstration proves that the model is feasible and efficient.

In order to improve the speed of auto-generating test paper system, this paper proposes an intelligent algorithm to generating test paper based on genetic algorithm, and provides a set of schemes of making paper of different degree of difficulties display in normal distribution. The algorithm adopts a new decimal system of subsection code, improves the traditional method of initializing the population and optimizes course of search. The experiment proves that this algorithm is superior to those available in the literature and more practical.

In order to predict the new real estate price, the factors affecting the real estate price were established by carrying out a survey of recently completed projects in the local environment. And a fuzzy neural network prediction model was developed. The model predicts the appropriate price for a new real estate by learning from historical data on the correlations between various factors that influence the prices of real estates and the actual selling prices. The experimental result shows that the fuzzy neural network prediction model has strong function approximation ability and is suitable for real estate price prediction depending on the quality of the available data.

For process algebra as a major branch of abstract description languages, almost no work directly exploits the structural symmetry in it. In this paper, a process is defined as an algebraic structure. We define a notion of symmetry for processes based on permutation groups. Given a process and a permutation group over it, we give the quotient process of this process and show that the quotient process is bisimulation equivalent to the original process. Moreover, we present the algorithm for this structural reduction.

Due to the flood of pornographic web sites on the internet, an effective web filtering system is essential. Web filtering has become one of the important techniques to handle and filter inappropriate information on the web. In this paper, we introduce a web filtering system based on contents. The system uses a probabilistic text classifier to filter pornographic information on the WWW. We focus initially only on Thai and English language web sites. The first process is to parse the web sites collection to extract unique words and to reduce stop-words. Afterwards, these features are transformed into a structurized “bag of words”. The next process is calculating the probabilities of each category in the naïve bayes classifier (as a pornographic web filter). Finally, we have implemented and experimented on our techniques. After testing by the F-measure, the experimental results of our system show high accuracy. This demonstrates that naïve bayes can provide more effectiveness for web filtering based on text content.

Traditional distance-based outlier detection algorithms neglect the local distribution feature of data and are sensitive to the selection of the parameters’ value. The paper proposes a novel outlier detection algorithm based on the even distribution pattern. The algorithm takes data’s local distribution situation into consideration during detecting outliers and adopts only one parameter, whose appropriate value can be automatically decided in the mining process. The paper also explores the way for the algorithm to handle large data set. The experimental results show the effectiveness and robustness of the proposed algorithm for the multi-type real-life data sets.

The existing approaches to mining episode and sequence have the disadvantage to the re-mine. Re-mine means that data are mined many times with various mining parameter. In order to solve the problem in the re-mine, the conception of Episode-time-table and re-mime is put forward. A novel approach to mining frequent episodes from event sequence is proposed, which is called CMFE (i.e., Continuable Mining Frequent Episodes). An Episode-time-table is corresponding with an episode pattern. The Episode-time-table records the time when every corresponding episode starts and ends. A new Episode-time-table is generated by joining two appropriate Episode-time-tables. The amount of records in the Episode-time-table is the number of episodes occurring in event sequence. Because the intermediate results, i.e. Episode-time-tables, is stored in the database, and the measure of joining two appropriate Episode-time-tables is taken to test candidate episode, CMFE greatly reduces both efforts of candidate episode test and the difficulties with mining long episodes to a certain extent. As the event sequence is re-mined, the Episode-time-tables are analyzed with various mining parameter to more valuable information. Re-mine needs not to scan original event sequence and retrench much required resource and time.

The study on streaming data is one of the hot topics among the database around the world recently. Now, we have many useful methods when we query in this kinds of new data. The sliding window model is a very important one. Some arithmetic using this model can help us finding the frequent items in data steams. However, most of them just considered the frequent times of each item, not considered the time span they arrive or the change pattern of data. In the paper, we introduce a new strategy using sliding window for finding some interesting patterns of steaming data.

Performance of ants clustering is very sensitive to the key parameters in ants algorithm, this paper analyses how to adjust key parameters adaptively to improve ants clustering. It also brings forward a improved method making use of pheromone and ant’s local memory to direct ant to move effectively, whereas ants are moving randomly after picking up or dropping an object in recent ants clustering algorithm. As the parallel nature of ants cluster, we apply AACA in web mining to mine correlative page set and make it suitable to dispose dynamic and large quantity of web log.

In this paper, we study directed graph combined with the reduction algorithms based on the principle of discernibility matrix. By investigating the discernibility set applied to directed graph, a new function opr( ) is given. The function focuses on the computing of the attribute set gotten from the discernibility matrix, so as to get a optimized set. Hence, we propose a better complete algorithm for attribute reduction based on the function opr( ). Finally we gave the proof of its completeness.

In order to overcome the shortcoming of general rough set theory, the elementary concept of tolerance rough set theory is proposed in this paper, and is used to build objects’ tolerance relations that can correctly classify objects in system. We use genetic algorithms to search for the optimal thresholds, then construct special matrix for attributes and objects. We can get the relations among attributes, objects and relative absorbent set of objects in detail. The method of using tolerance rough sets reduces the qualitative processing and improves the validness of knowledge reduction. Finally, we present examples, illustrating our approach in the paper.

Search engines based on content similarity algorithm prevail in Web retrieval field, but they return large volume of Web pages, and therefore, make the user wait longer time. Search engines based on Web hyperlink structure analysis algorithm play a significant role in improving the precision of searching results, but they prone to cause the polarization towards Web pages. To fix these drawbacks, this paper proposes a new searching algorithm that combines the Web content analysis and Web linkage analysis. Based on the PageRank, we introduce an accelerating factor, which can reflect the changeable speed for the importance of a Webpage. Moreover, according to the importance of the two algorithms described above, we present two parameters expressing their weight values. The experiments show that the new integrated algorithm is improved.

The mining of cube gradient is an important research problem in the data mining community. It is an extension of the traditional association rules mining in the data cube and has broad applications. The existing works mainly explore for the studies of cube gradient mining algorithms. There are fewer studies on the interactive tool of cube gradient mining. In this paper, an online cube gradient mining tool, OLGM, is presented. The three tires system architecture of OLGM based on the browser/server model is presented. The key implementation techniques of OLGM based on the dmGQL gradient query language and condensed data cube techniques are introduced. The system features show that OLGM is effective and extensible.

In industrial, scientific and commercial fields, geographically dispersed large data sets are necessary to be analyzed to get a lot of useful information by using distributed data mining (DDM). However, DDM now confronts many new challenges such as seamless-cooperation which can not easily be solved on current distributed computing environment. As the latest development of distributed computing technology, the gird computing may play a key role in solving these challenges. In this paper, the shortcoming of current existing DDM systems is showed, and a service-oriented architecture of DDM on the grid is proposed. The mining algorithm, distributed data sets and computing power are abstracted as web service resource (WS-Resource), which can cooperate to perform DDM as required dynamically. In addition, the paper describes the mining process on the proposed framework in detail.

Web crawler is the core component of WWW search engine. This paper discusses a topic-specific intelligent Web crawling algorithm CA(C&S) based on Web content and structure mining. The algorithm takes full advantage of the characteristics of the neural network and introduces the reinforcement learning to judge the relativity between the crawled page and the topic. When calculating the correlation, only extract the important tags of HTML makeup of the Web page, to analyze the content and structure of the page. The experiments prove that the algorithm improves the efficiency and accuracy enormously.

The rapid development of the microarray technology brings about a great challenge to conventional clustering methods. The sparsity of data (sample), the high dimensionality of feature (gene) space and many irrelevant or redundant features all make it difficult to find correct clusters in gene expression data by using conventional clustering methods directly. In this paper, we present CIS, an algorithm for clustering biological samples using gene expression microarray data. Different from other approaches, CIS iterates between two processes, reclustering genes (not filtering genes) and clustering samples. Inspired by the policy of refining progressively, CIS repeatedly partition the set of initial genes with the new-generated sample clusters as features, and then partition samples with the new-generated gene clusters as features over again to identify the significant sample clusters and relevant genes. The method is applied to two gene microarray data sets, on colon cancer and leukemia. The experiment result show that CIS works well on both two datasets. We partition the two sample sets by eight and twentynine genes respectively, thus both acquire the accuracy about 90%. All these indicate that the CIS might be a promising approach for gene expression data analysis when domain knowledge is absent.

As the Web grows, more and more data have become “hidden” or “deepen”. Web crawlers should not only have the ability of fetching the Publicly Indexable Web (PIW), but also be able to plunge into the Hidden Web to search for more useful information. In this paper, an approach for exploring and identifying the domain-specific search interfaces by using SVM classification scheme is presented. The method of integrating domain-specific search interface (DSI) into the topical crawling system is also introduced. This work is based on our observation of the conciseness and representative characteristics of DSI. Though intuitive and apparent this approach is, it seems more preferable for identifying DSIs. The experimental results show that such a feasible and practical way can achieve good performance.

Dominance-based Rough Set Approach (DRSA) is an extension of classical Rough Set Theory (RST). Approximation of class unions is a very important approach in DRSA. Aiming at the disadvantage of the classical method we presented a new methodology for approximation of class unions based on dominance-matrix. It only needs to calculate the dominance-matrix. Thus it greatly simplified the process and reduced the complexity by about k times. Besides it is intuitive and efficient. The example illustrates its feasibility and efficiency.

Application of information technologies (IT) in the field of audit is worth studying. Chinese data-oriented IT audit is researched in this paper. Firstly, the conditions of IT audit used in domestic and overseas are analyzed. Then, the main steps of data-oriented IT audit used in China are presented, including: data acquisition, data cleaning and data processing. And data processing method based on outlier detection and visualization data processing methods are proposed. Then, the popular data-oriented IT audit software developed in China are given. Finally, data-oriented IT audit used in Chinese online audit are introduced, and the differences of data acquisition and data processing methods between online audit and single computer audit are analyzed.

Euler Histogram is one of useful methods to estimate the number of the spatial objects that intersect with the query window. Several cost models of spatial queries are based on it. In this paper, we present a new mothed SP_Histogram to estimate the selectivity of spatial object. In SP_Histogram, spatial object is approximated by Simple Polygon, and it have higher calculative accurate than Euler Histogram and Closed_Euler Histogram. The correctness of SP_Histogram is also proved.

In order to study the process of a fuzzy concept translated among multi-agents, the concept of rough communication of fuzzy concept is proposed. In a rough communication of fuzzy concept, the problem of uncertainty exists, for each agent has a different language and cannot provide precise communication to each other. The quantization analysis for the problem of information loss in a rough communication of fuzzy concept is discussed. By using some concepts, such as α —rough communication cut, the relation theorem between rough communication of fuzzy concept and that of classical concept is obtained.

In data mining, clustering is used to discover groups and identify interesting distribution in the underlying data. We propose a new clustering algorithm called CMR that is more robust to outliers, and identifies clusters having non-spherical shapes and wide variances in size. CMR achieves those by representing each cluster by multiple representatives. Having more than one representative per cluster allows CMR to adjust well to the geometry of non-spherical shapes. CMR use atomic-clustering algorithm to get atomic clusters on which the hierarchical merging algorithm is performed. CMR is a polynomial-time clustering algorithm, and so it facilitates the clustering of a very large data set.

The techniques of workflow mining can support the design of workflow. It is a challenge to exploit more information in a set of log. In practice, many business processes can vary with the change of time and environment. This paper presents a method to mine and describe the dynamic information in process log. A new conception named dynamic workflow net (DWF-net) was proposed, which evolves from workflow net (WF-net). DWF-net has new attributes to characterize the dynamic factors in workflow models. This paper also presents the idea and steps of the algorithm that can generate a DWF-net from a set of workflow log. The prototype and experiments have proved that the algorithm is effective.

A robust and efficient decision tree — R-C4.5 and its simplified version R-C4.5s were proposed in order to enhance the interpretability of test attribute selection measure, reduce the numbers of insignificant or empty branches and avoid the appearance of over fitting. This model is based on C4.5 and improved on attribute selection and partition methods. R-C4.5 combines branches which have high entropies, because these branches have poor classification effect in divide-and-conquer process. The sensitivity of R-C4.5 to missing data was also studied to find out the influence of missing data on R-C4.5. The results show that R-C4.5 improves the predictive accuracy and robustness. It constructs decision trees smaller and less sensitive to missing data than C4.5 does.

The emerging electricity market of imperfect competition makes it possible for each power supplier to gain more profits through strategic bidding. However, no matter what kind of bidding strategy, it is essential to gain mass data and analyze them. Therefore, the data warehouse based on Web is applied to bidding strategy for power supplier in this paper. By this way, it would supply reasonable information to the decision maker. In order to take full advantage of web-based data warehouse, the author deeply analyzes each level in the classic triple-level structure of it. By introducing the software Agent technology to every part of the structure, the author proposes a web data warehouse system structure based on the Agent to solve the shortages of the web-based data warehouse system that traditional method can’t solve. To the important issue that how to integrate the useful web data to data warehouse, the method is proposed to transfer the HTML page into XML data source combined with Agent technology. So, the power supplier could make the best of network resource, and transfer them into own data warehouse for building the optimal bidding strategy. Additionally, the applications of Agent on the application server layer and on the data warehouse system are analyzed respectively. According to this method, the feasible and efficient web-based data warehouse could be constituted for the power supplier.

Web systems, due to their peculiar nature, can not be just developed as conventional software projects. This encourages us to view Web systems as special kind of software systems; and therefore, to blindly apply existing models to the representation of Web systems persuades the developers to overlook the peculiarities of these Web systems, and hence leading to inappropriate solutions. We need to understand how these limitations may be circumvented by appropriately supplementing (or replacing where necessary) the existing models. In our previous work we have proposed a process Framework Web OPEN (Object-oriented Process, Environment, and Notation) Integrated [1] where web systems can be developed in an engineered manner. In this paper we extend our previous work by proposing new Tasks, Subtasks and Techniques, and then we present how these additional Tasks will address the peculiar issues of web systems.

Security plays an important role in current information systems. When developing an information system, one will deploy appropriate security policies from requirements and enforce them to the system. It’s strongly necessary to verify whether these security policies are successfully enforced to the system. This paper proposes a methodology, based on model checking, to verify the information flow security policies enforced on various information systems. Our method, which is pretty general and practical, can be used to verify information flow policies from both the system’s model layer and its implementation layer. The Information Flow Graph (IFG) is presented as a theoretical verification model for information systems. Then we use a general coding method to encode the states and transitions. Thus the verification may be done using practical model checkers such as Spin. In special, we apply our model to the verification of secure operating systems equipped with Mandatory Access Control (MAC) framework.

So far most research work done on software reverse engineering is mainly focused on the legacy systems, which are built on obsolete technologies, but contain a lot of information and knowledge about the certain application domain. However, software based on more recent technologies, such as component-based, also raises many issues during comprehension and evolution. Especially, the component models like in J2EE world are based on many technical aspects that make them difficult to understand and use. Thus, the evolution of component-based software becomes an emerging issue in reverse engineering area. In this paper we propose an idea to integrate reverse engineering with more recent systems based on the new technology. component-based, not only restricted to the legacy system. We also extend the existing graph based approaches to reverse engineering such systems. In our approach, the directed graph describes and manages dependencies among the components in the system based on the information extracted from source code, deployment descriptors and configuration files. With this component dependency graph, it helps developers to understand the system better, and ease the succeeding phases of reengineering the system.

Though contributing to the significant improvement of software quality, the application of formal methods is limited by rigorous constraints, especially in large systems. As an alternative, partially introducing formal methods avoids the high cost for entire projects. During software evolution, software quality metrics of the existing system can be useful heuristic criteria for selecting worthwhile parts for formal treatment. We first select extensive candidates with high failure rate or high complexity, and then eliminate those with low customer problems, minimal development or maintenance efforts, thereby get a subset of the system to which applying formal development techniques can be most likely payable. Case studies with three real-world systems are presented to illustrate our approach.

This paper introduces a novel sketch-based CSCW (Computer Supported Collaborative Works) system for UML design. Being different from traditional CSCW and CAD systems, our system uses free hand sketch as user interface, which will fully exploit the advantages of sketch — natural, convenient, and more capable of handling brainstorm. When using sketch as interface in CSCW systems, two problems have to be solved first: (1) how to store, edit and exchange sketch-based document consistently; (2) how to visualize sketch-based document. To solve the first problem, we propose a new document format SKETCH, which is able to represent sketches and related semantics. To solve the second problem, we propose an ontology-based conceptual model which is helpful of recognizing sketches and providing uniform visualization of sketches. To demonstrate our ideas, we have developed a prototype system that can support simple UML Class Diagram design in collaborative environment. It shows us good results, but still, more works are going to be done in the future.

Requirement engineering covers all activities related to elicitation, specification and validation of business processes, in which data origin relationship is critical and important to elicit and describe user requirement. This paper proposes a new requirement elicitation and analysis method: Subject-Predicate-Object Method (SPOM) to mine and analyze data origin relationship, using Form Flow Diagram (FFD) as starting step and analyzing based on Data Origin Diagram (DOD), and presents the definitions, design steps and key steps analysis of FFD and DOD, with which preliminary data, data model and static data description of user requirement can be described clearly in requirement analysis phase. This method has been applied in one ERP project to mine and analyze data origin relationship of product storage business.

Existing software reuse strategies focus on reusing program codes or binary codes. However, reusability is not universal property of program codes, software reuse has never achieved its full potential. Requirements within the same domain are more likely to be similar than the software components. Requirements reuse can provide an enhanced opportunity for reusing software components. Traceability and understandability of requirements model are critical for requirements reuse. Current researches concentrate on unstructured and informal requirements modeling. It is hard for the customer and the developer to trace and to understand the requirements model. In this paper, we construct requirements model by architecture based modeling approach, and the relationships between components are defined formally. Requirements model consists of organization patterns and requirements partition patterns, the traceability between problem space and software requirements is established as the side effect of requirements modeling. The constructed requirements model is traceable, maintainable, and understandable, hence reusable.

The practice of software reengineering has not been fully exploited industrially in significant ways. In this paper, we will report our experience with the architecture based reengineering on a Network Planning System (NPS). Our approach for effective reengineering is focusing on architecture recovery and component reuse. The new architecture is built based on the recovered architecture, which is augmented with new architectural requirements, correcting old design flaws, and incorporating design patterns. The presented approach is evaluated with the reengineering of the NPS where performance, productivity, code quality and extensibility are compared with the conventional approach. The experiments show that our strategies are very successful.

The paper proposes an XML format for Object-Z specification language. We describe several markup format of Object-Z language and compare some existing works, the problems and the issues that arose, and propose a new XML markup format of Object-Z specification language. The new proposal considered the particular requirement of Object-Z specification animation and demonstrates the feasibility and validity by an example.

Process change extremely affects software quality and development efficiency, and process dynamic change realizes flexible enactment. Supporting dynamic change and flexibility is required when PSEE (Process-centered Software Engineering Environment) is employed. However, there are few researches on change management involved completed activities, which is regular in software process because of its iteration. In this paper, we consider change management in software process. Firstly we analyze reasons and corresponding strategies of dynamic change; then propose the criteria of “Correct Termination” to judge the rationality of software process dynamic change; finally, we analyze the cost of dynamic change and present a research prototype — SPEFE, which supports process dynamic change.

A novel cut-and-paste editing method over 3D triangular meshes is proposed in this paper. This method pastes the source mesh onto the target mesh directly. The source feature mesh is transferred to the position where it is to be pasted through the corresponding planes. The planes are created through the least squares approximation. Then, the boundary vertices of the source feature mesh are connected with target mesh. This method is applicable to source geometry features with non-zero genus. This algorithm can be performed at interactive rates and produced less distortion and natural results.

Software outsourcing has become more and more popular in recent years, and shows a non-reversible trend in the IT area. At the very beginning, the main purpose of software outsourcing is targeted in cost saving. But now it has turned into quality improvement and competence focus where software process will have crucial influence. This paper will analyze some experience of Indian software outsourcing industry, and present a new process for software outsourcing in China. The process is based on SITO Model and involves the Software Process Improvement (SPI). It has already been adopted in State Street Zhejiang University Technology Center (SSZUTC) as a proven record in process improvement.

In the component-based software development (CBSD) process, developers need an effective selecting method to search eligible component. That is how to choose the most appropriate and suited component from component library effectively. This paper proposed a new approach of evaluating component on the basis of linear comprehensive approach. We analyze a set of factors and their weights, and put forward the method of arranging components in an order after classifying in the component library. Our experiment shows the approach is better than other approaches such as subjective mind and simple linear weighting method.

Component-Based Software Development (CBSD) is gaining substantial interest in the software engineering community. Currently, components delivered by component providers only include specifications of the interfaces. Without source code, many traditional software testing methods can not be applied to component testing. We have developed an approach for determining the quality of components using a set of black-box analyses based on function domain partition and defined a component testing process in the paper. This approach will help component users test and understand components, thus provide much information useful for choosing components.

An important emerging requirement for distributed systems is adaptive QoS, which necessitates more flexible system infrastructure components that can adapt robustly to dynamic end-to-end changes in application requirements and environmental conditions. This paper shows how to add a characteristic of dynamic configuration, from the perspective of fault tolerance QoS, to the fault tolerant object management framework based on middleware. Three specific technical challenges in satisfying the requirements are: 1) perceive the changes of environment; 2) define the trigger for reconfiguration; 3) reconfigure the resources if the trigger satisfied. We define dynamic reconfiguration as the relation among fault tolerance properties, description of computing environment changes and dynamic adjustment algorithm, Based on this definition the management framework for dynamic reconfiguration is built. In which, the introduction of reflection model and publish/subscribe model makes the acquirement and notification of information more convenient and flexible, and the adoption of policy customization mechanism makes reconfiguration enforced more effectively. The framework presented in this paper has been validated in a prototype.

Component-based software development approach is one of the most promising solutions for the emerging high development cost, low productivity, unmanageable software quality and high risk. This approach, however, encounters some problems about unseparated crosscutting concerns that are easy to lead to the code-tangling and code-scattering. Aspect-oriented programming enables the handling of crosscutting concerns and implements the separation of concerns. It will help to understand the software system better and strengthen the adaptability, maintainability and reusability of the final software. This paper works on the combination between CBSD and AOP. The combinatorial semantic constraint is described in theory, and then proposes a framework and sets up its supported environment. Taken book management system as an example, the framework is proved to be an effective way on the road of the combination between CBSD and AOP.

With the increase of size and complexity of software system, more and more developers focus on component-based system developing. For testing these systems, this paper presents an approach, which is called connection pattern based testing method. And a formal notation for describing the connection and components--TADL (ADL for testing) is presented. Details of TADL are discussed and an application is presented to modeling Microsoft Speech SDK.

Design patterns are widely used to achieve large-scale reuse by capturing successful practice in software development, but their implementations still remain redundant and burdensome. We present a category theoretic approach to mechanize the refinement process from design pattern templates to executable programs. A pattern template and its “standard” refinement path, which are expressed with theory-based specifications and morphisms, can be saved in a library. When encountering another problem where the same pattern template needs to be applied, instead of performing a new refinement, we can work out its provably correct implementation using category theoretic operations.

Model-checking is an automatic technology for the verification of the temporal property of concurrent or real-time systems through explicit state exploration or implicit fix point computation. The systems are usually described by the automata and the properties of the systems are expressed by the temporal logic formula. The correctness of the real-time systems not only depends on logic result but also on the timing constraints. For the analysis of the real-time systems, we use timed automata and the LCTL which is the temporal extension of branching-time logic CTL to express the real-time systems and its properties respectively. Choosing dense domain instead of discrete domain may blow up the complexity of model-checking problem; in terms of Alur’s method, we can translate timed automata into region automata as a solution.

With the database becoming ever-larger, the architecture is developing continually in order that the data and resource can be shared. There are many problems exiting in Distributed Database Based on C/S. After analyzing the advantage and disadvantage of single-tiered, double-tiered and three-tiered architecture, this paper proposes a solution of multi-tiered architecture. Finally, this paper gives an example of the multi-tiered architecture to demonstrate the advantage of it.

Specification-based software development can allow the development, verification and maintenance of software at the specification level and can synthesis provably correct code. However, formal specification technology is hard to apply to the industrial development of software. UML has been used widely in the CASE-based software development. But, UML is a semiformal language and lacks formal semantics to express its meaning precisely. The paper proposes an approach for the formalization of UML statecharts mapped to RSL -a specification language based on coalgebra. The purpose is to bridge between UML and formal specification language.

The paper presents a kind of implementation method of the mobile middleware based on BREW mobile device, which combines the BREW with MICO according to the characteristic of the mobile device and implements the interconnection, intercommunication and interoperation between mobile data application and Internet and desk area networks.

Concurrent Engineering is a newly integrated, and collateral systemic method which is used to design the products and other interrelated processes of these products. Although foremost put forward in the industrial manufacturing field, the concurrent engineering idea here will apply in the software engineering field. Based on the features of concurrent software engineering, we produce the X-Petri net (XPN-Extended Petri net), and try to use it to describe the software concurrent process model that we have built. At the same time, based on the hypothesis of the normal distribution, we work out the performance equivalence formulas of three basic execution structures in the software concurrent process, that is, sequential, concurrent and cyclical, and give proofs of these formulas.

One of the major problems associated with CBSD is the difficulty in searching and retrieving reusable components that meet the requirement at hand. A similar matching-based approach to component retrieval is proposed as a solution to this problem. This approach makes use of a hierarchy semantic-based component representation that exactly expresses semantic features of components. A multi-dimensional component library is developed based on virtual components to implement automatic categorization of components. An experiment is provided to illustrate the performance of the proposed approach.

Software process improvement has gained added significance in both industry and academe over the last few years. Compared to multitudinous definitions of software product quality, the research on software process quality is almost blank. Relying on statistical process control (SPC) techniques, this paper defines Total Quality and Partial Quality of software processes, summarizes seven attributes of software process quality, divides assignable causes into process-inner assignable causes and process-outer assignable causes, and uses Shewhart control charts and Select Cause Control (SCC) charts to analyze the quality of a specific process. Experiences show that the work in the paper can strengthen the understandability of software process improvement, and establish the relationship between software process quality and software product quality that is not well addressed in popular process models, such as CMM.

First of all, the typically dynamic workflow systems are introduced, and their main defects that are not able to decide automatically the dependences among data in workflows are analyzed. The reasons resulting in the problem are lack of effective modeling approach and constructing algorithm of dynamic workflow, which are our main research work in this paper. Here, the Petri-Net is adopted to model the dynamic workflows. Based on the model a constructing algorithm of workflow is proposed, and the complexity of the algorithm is analyzed. Lastly, by a computing example the modeling approach and constructing algorithm of dynamic workflow based on Petri-Net are proved to be practical, effective and feasible.

Network processor is used to compose the data path for various network application systems. It is difficult to develop such system due to the complex hardware architecture and strict performance requirement. However, component technology is a promising approach for embedded system. In this paper, we (1) identify the generic model of computation in data plane of various application systems based on network processor; (2) define different components to fit in with the computational model and function structure; (3) discuss requirements for software component model and framework; And in the end, a specific component model is proposed in the framework of fast data path. The proposed component concepts are suitable for behavior specification, system building, and for performance prediction and analysis. The discussed component models and frameworks are based on IXP2400, while the basic concepts could also be applicable to other network processors.

The paper presents a design and its implementation on a physically based animation of a lovely squirrel mascot playing Chinese Tai Chi. The work result was from a project aiming at an application of motion tracking technology to generate physically based animation for the Macao 2005 East Asian Games, in which the mascot called Pak Pak will play Tai Chi inside a virtually constructed Macao Sports Dome. This paper describes the modeling of the virtual character Pak Pak as well as some issues in applying the motion data captured to animate the character. Some solutions on modeling the complex part of squirrel shape and on the solution of fitness problem of human data applied to squirrel character are also proposed. We will go through both the ideas and technical detail on the solutions. A complete animation on the fly-over of the sports dome as well as the lovely Pak Pak playing Chinese Tai Chi will be presented along with the paper presentation.

Binocular stereo vision is one of important branches of stereo vision. This paper first introduces a binocular stereo vision model. Then focuses on the stereo effects that can be seen through the model (i.e. the model’s properties): the method to calculate the coordinates of stereo pairs is given, properties of parallax are presented, and the stereo effects when eyes moving in the same plane which parallels with the screen are also discussed. Finally, conclusions drawn in the paper are proved by experiments. The results show that if the distance between observer and screen do not change, he won’t feel the objects move away or come close when he shakes his head slightly, but he will feel the whole image is swaying. Though slight differences exist for different people, the phenomena each person sees accord with predictions we deduce from the model.

Triangulation from scattered points is one of the most important processes in 3D surface construction. Traditional methods generate surface with defects such as gaps, holes and self-intersecting polygons. Thus, additional reparations such as cutting and stitching methods have to be used to create a closed surface. In this paper, a novel method of rapid synthesis of closed manifold surfaces from scattered data is presented. Cluster analysis is first used to divide scattered points into different clusters. Then, each cluster is projected and triangulated separately. Finally, the patches of triangular meshes are glued step by step in a certain order to create a closed manifold surface. In considering that some scattered points could be partitioned into the wrong patch, aided planes are used to help adding points into the correct point sets. Moreover, the same method can be used to accelerate the process with huge number of scattered points. Experiments demonstrate how this approach can relatively easily produce a good visual quality of closed manifold surfaces.

This paper presents a new method of constructing statistical shape model for face shape modeling and this model is capable of generating shape variation. Traditional statistical shape models are constructed by approximating the distribution as a Gaussian distribution and the shape variation is represented as a linear function, which can not adequately represent non-linear deformation. Here we analyze the positions of points set and find that fuzziness and randomness in the shape deformation can not be denied. The unified model, based on cloud model, is presented to describe the face shape and its deformation. We describe the model in detail and give examples of the resulting model.

A block placement representation called sequence pair was introduced by Murata. Many block placement algorithms that are based on sequence pairs use simulated annealing where the generation and evaluation of a large number of sequence pairs is required. This paper presents a new approach to evaluate a sequence pair based on computing longest common subsequence in a pair of weighted sequences. We present a very simple and efficient algorithm which has complexity lower than O(M^1.5), it is a significant improvement compared to the original O(M²) algorithm[1]. Experimental results show that our approach can estimate the area of a placement faster.

Automatic recognition of Chinese personal name is important and difficulty for unknown words recognition. We obtain three groups of data from tagged corpus, that is p(i), Q(i), R(i), representing the probability of surname, the probability of the word of surname, the probability of the surname word being a real surname. We proposed an approach of automatic recognition of Chinese personal name based on the priority of surname, and provided some algorithms according to the method. The system of automatic recognition of Chinese personal name based on this paper reaches high recall rate (80.61%) and precision (89.27%), having application.

A method for recognizing motion objects in surveillance video is presented. Firstly, according to fuzzy theory, a fuzzy feature vectors (FFVs) extracting algorithm is presented, which describes the object features by membership degree. Finally, with a “one-against-rest” classifying strategy, the multi-class pattern recognizing problem of motion objects is solved by using a multi-level two-class support-vector-machine (SVM) classifier to classify and recognize. So, motion objects in videos are classified into four patterns which are named as “people”, “vehicle”, “motorcycle”, and “bicycle with man”. The experimental results show that the method presented in this paper is feasible and good. It can correctly recognize each motion object even in the situation of the object has been scaled and distorted.

Fingerprint matching is a key technique of automatic fingerprint recognition system. The result of point pattern matching is accurate while its repeated search for maximum likelihood pairs are very time-consuming. In this paper, we proposed a fast complex fingerprint matching method to solve this problem. The innovative method constructs rotation-and-translation-invariant feature vectors to describe local information of minutiae sets and shifts the matching style from point pattern to feature vector. If the decision is uncertain by this case, then it estimates rotation and translation parameters from previous matched pairs and returns to point matching. Since this method takes the similarity and consistency of local deformation into account, it is not only much faster, but also lower false accepted than traditional approaches. The test on FVC2004 data sets shows that the equal error rate (EER) is 2.97%, and the mean matching time is 10.3 milliseconds, variance is 2.03, superior to most traditional methods. When minutiae number is among 15-40, the EER reaches to 2.14%.

Medical image segmentation is one of the most productive research areas in medical image processing. The region growing algorithm is the most popular and efficient approach in this field. The key point of this method is to define the homogeneity criterion which is difficult to clearly describe and highly relied to the experience of the researcher. In this paper, we proposed an adaptive region growing algorithm which used gaussian mixture model (GMM) to describe the homogeneity and image shape properties of the local region in the medical image. Firstly, some statistical parameters are estimated by investigating the characteristics in the local region. Then these parameters are obtained by analyzing the statistical information using certain cluster algorithm. Lastly the growing parameters are automatically computed and applied in the region growing algorithm. To reduce the affection of the noise, we employ the anisotropic and multi-slices gaussian filter in the image processing. This method was tested with many CT and MR images. The experimental results show the approach is reliable and efficient.

Daubechies orthogonal wavelet transform is used to preprocess the face image, resulting in its four sub-images belonging to different frequency bands. The four sub-images were analyzed by the PCA method respectively, and the nearest neighbor classifier is used to recognize them. This makes correlation reduce and makes difference rise between the four groups of sort results. Moreover, all of or part of them can be combined according to group consensus index. Experiments show that so far as the PCA analysis is concerned, there is a certain complementarily between the sort results of the four sub-images. The sort performance can be improved.

Because the 3D sights is very complex, it is difficult to realize the simulation and visualization of 3D sights on internet. We design and develop a system of B/S architecture for interactively browsing the 3D sights on the Web. To support the simulation of 3D sights with high performance, an extended-relational logical model of 3D sights is designed, based on this model, the sights data can be organized and managed effectively by Oracle spatial. In addition, a specification for transporting sights data between Web server and browser is defined with XML, Additionally, an effective strategy for sights data schedule is designed to support the real-time rendering of 3D sights. With our system, user can interactively browse complex 3D sights on the node of internet.

Fuzzy clustering is playing a more and more important role in text clustering because of text diversity and abundance. As the most popular fuzzy clustering algorithm, FCM, however, is rather sensitive in its initial clustering centers. This paper presents a new GA-based FCM approach (GFCM for short) to overcome this drawback, which optimizes the initial clustering centers of FCM with the global searching ability of GA. Related operators are improved to enhance clustering quality and accelerate searching process. Besides experiment results not only prove its feasibility but also reveal more effective performance, compared with previous FCM algorithm.

Modeling of 3D highway integrated with terrain model was firstly discussed in this paper, and a new method of judging the maximum angle was used to eliminate the distorted triangles on the boundary of model. It is a fundamental issue for visualization and analysis on 3D GIS at a rapid speed. To strip-distributed terrain data, this paper proposed a new dynamic multi-resolution model. Different from other models/algorithms, this algorithm need not pre-partition the model regularly and can achieve multi-resolution representation in one tile. In addition, such a condition was considered in the approach that only resolution level of parts of triangles will change in adjacent frames when roaming. That is, the resolution of parts of triangles which locate in the joints of adjacent resolution levels will change. Therefore, the speed of algorithm can be improved. Finally, a software prototype “VRHighWay”, developed by VC++ 6.0 and OpenGL, was introduced. The experimental results demonstrate that proposed method acquires better performance in terms of accuracy for the multi-resolution representation of terrains with roads embedded.