International Journal of Software Engineering and Knowledge Engineering

Several methodologies have been developed to enhance the software life cycle of knowledge-based systems by emphasizing on the use of both prototypes and specifications. However, these methodologies focus on the development phase of knowledge-based systems. The roles of prototypes and specifications in the maintenance phase has not been fully explored. Because a suitable problem specification for a knowledge-based system is often difficult to acquire, validating changes to non-executable solution specification during the maintenance phase can be a problem. To address this, we propose an alternative paradigm in which the prototype complements the specification throughout the life cycle. The traceability between them is facilitated by organizing both types of artifacts using a common functional decomposition structure. Based on our task-based specification methodology (TBSM), we have also developed a knowledge engineering tool (called TAME) to facilitate the acquisition and the organization of the specification and the prototype. The proposed methodology and the tool together can thus enhance the verification, validation, and the maintenance of knowledge-based systems through their life cycles.

Expert system technology is being applied to complex problem-solving tasks in many domains; concerns have naturally arisen as to the reliability of these systems. As with any software, attention to quality and reliability must be paid throughout development of an expert system. Validation and verification techniques must be employed, and effective use of such techniques depends upon the existence of specifications which state not only what the system must do, but also how it must do it. This article presents practical and sound techniques for performing specification, verification and validation activities for an expert system. Tools are described which assist the developer in performing verification and validation. It is argued that these tools and techniques form the backbone of a powerful and practical expert system development methodology. The general aim of the article is to demonstrate that formal processes can be applied to assure the reliability of an expert system, and that use of such processes is entirely practical, given the present state-of-the-art in software engineering and expert systems. The article is illustrated with examples from the development of a real expert system application in a medical domain.

Knowledge based systems are often used to replace humans in solving problems for which only heuristic knowledge on the solution is available. However, there are also important application areas where nonheuristic knowledge is available e.g. in technical documents but where efficient use of the knowledge is impossible without the techniques provided by artificial intelligence. High dependability of these kinds of applications can be achieved if domain knowledge can be represented in a language providing both adequate representational constructs and the required level of formality. In addition, the language should be supported by powerful tools assisting in the verification process.

Knowledge Based Systems, despite the different technology employed, are still nothing more than a computer program. Unfortunately, quite a few people building knowledge based systems seem to ignore the many good programming practices that have evolved over the years for producing traditional computer programs. What we need is a framework for the modelling of the KBSs development.

In our work, it is claimed that these requirements can be met by utilizing and combining ideas from control engineering, software engineering and artificial intelligence.

In this paper we develop a software reliability model for Artificial Intelligence (AI) programs. We show that conventional software reliability models must be modified to incorporate certain special characteristics of AI programs, such as (1) failures due to intrinsic faults, e.g., limitations due to heuristics and other basic AI techniques, (2) fuzzy correctness criterion, i.e., difficulty in accurately classifying the output of some AI programs as correct or incorrect, (3) planning-time versus execution-time tradeoffs, and (4) reliability growth due to an evolving knowledge base. We illustrate the approach by modifying the Musa-Okumoto software reliability growth model to incorporate failures due to intrinsic faults and to accept fuzzy failure data. The utility of the model is exemplified with a robot path-planning problem.

A Cooperative Response system, which can detect and correct users’ misconceptions exhibited in database retrieval, is one of the most promising ways for applying the knowledge engineering approach to traditional database query systems. Misconceptions are generally regarded as users’ false beliefs about a database and are considered as a main source of failures in the database retrieval. How to handle users’ false beliefs has recently become an important point to overcome in advanced database front-end development.

While many efforts have been made to develop cooperative response systems, most of the work has placed emphasis on the detection of a user’s false beliefs revealed in a single query. No attention has been paid to handle users’ inconsistent beliefs which would be revealed in a sequence of contextual queries. In this paper, we propose a user-beliefs-modeling approach for detecting the inconsistent beliefs exhibited in two contextual queries. As an extension of the existing cooperative response systems, the system proposed in this paper can generate informative responses to correct users’ inconsistent beliefs.

Researchers who create software production environments face considerable problems. Software production environments are large systems that are very costly to develop. Furthermore, software production environments which support particular software engineering methods may not be applicable to a large number of software production projects. These conditions have formed a trend towards research into ways which will lessen the cost of developing software production environments. In particular, the trend has been towards the construction of meta-environments. In this paper, we attempt to categorize current meta-environment approaches. For each of the categories, we review research efforts which illustrate different approaches within that category. We conclude by presenting an emerging common thread of requirements which links this field together.