International Journal of Uncertainty, Fuzziness and Knowledge-Based Systems

This paper deals with the extraction of default rules from a database of examples. The proposed approach is based on a special kind of probability distributions, called "big-stepped probabilities", which are known to provide a semantics for non-monotonic reasoning. The rules which are learnt are genuine default rules, which could be used (under some conditions) in a non-monotonic reasoning system and can be encoded in possibilistic logic.

Many knowledge discovery tasks consist in mining databases. Nevertheless, there are cases in which a user is not allowed to access the database and can deal only with a provided fraction of knowledge. Still, the user hopes to find new interesting relationships. Surprisingly, a small number of patterns can be augmented into new knowledge so considerably that its analysis may become infeasible. In the article, we offer a method of inferring the concise lossless and sound representation of association rules in the form of maximal covering rules from a concise lossless representation of all derivable patterns. The respective algorithm is offered as well.

In this paper, we study the properties of flexible queries in the OLAP (On-Line Analytical Processing) framework, focusing on unary operators. For this purpose, we consider the model we have defined for fuzzy multidimensional databases. This model provides means to handle fuzzy data and flexible queries. The operators defined in this model are closed on the set of fuzzy hypercubes (hereafter cubes), which means that the result of each operator on a fuzzy cube is a cube. Thus, these operators can be nested into expressions. In this paper, the combination of several queries is investigated in order to study the possibility for the definition of an algebra to manipulate fuzzy cubes. This would provide a framework for query rewriting and, as a result, for query optimization.

In this paper, we define the concept of generalized rule for making classical deduction with imprecise data, stored both data and rules in a fuzzy relational database represented in the GEFRED model. We propose a way of measuring the imprecision related to the calculation of a fact based on the matching degree of the facts in the database and the facts calculated while expanding the rules. In order to achieve this, classical algorithms for deduction are not appropriated and we propose the modifications that have to be applied on a classical tuple-oriented algorithm in order to design a new algorithm for deducing from imprecise data with generalized rules.

Importance weighted averaging is a central information processing task in multicriteria decision problems of many kinds, such that selection, classification, object recognition, query answering, and information retrieval. These problems are characterized by a query, i.e., a set of importance weighted criteria, and a set of options queried. While each criterion determines a ranking of the options, the task of the averaging operator is essentially to aggregate these rankings into an overall ranking under the consideration of the criterion importance. We present a class of such operators, based on the power means, namely the Andness-directed Importance Weighted Averaging (AIWA) operators. The operators are equipped with an approximate andness measure allowing an easy, direct control of the andness in the unit interval. The aggregation behavior of the operators appears to be similar to that of importance weighted maximum entropy OWA operators. However, AIWA operators aggregates n arguments (criterion satisfaction values) in O(n) time as opposed to O(n log n) time for OWA operators. An interesting property provided by AIWA operators is decomposability, allowing us to consider new or improved criteria without recomputing with all arguments. Overall, the AIWA operators appear to be effective as andness controlled, importance weighted averaging operators, as well as easy to apply and computationally efficient.

Information retrieval is becoming increasingly concerned with resource selection and data fusion for distributed archives. In distributed information retrieval, a user submits a query to a broker, which determines a solution for how to yield a given number of documents from all available resources. In this paper, we present a multi-objective model for resource selection, in which four aspects: a document's relevance to the given query, time, monetary cost, and the chance of getting document duplicates from resources, are considered simultaneously. Some variants of this multi-objective model, aimed at achieving better implementation efficiency, are also proposed.

The Bayesian Network Retrieval Model is able to represent the main (in)dependence relationships between the terms from a document collection by means of a specific type of Bayesian network, namely a polytree. However, although the learning and propagation algorithms designed for this topology are very efficient, in collections with a very large number of terms, these two tasks might be very time-consuming. This paper shows how by reducing the size of the polytree, which will only comprise one subset of terms which are selected according to their retrieval quality, the performance of the model is maintained, whereas the efforts needed to learn and later propagate in the model are considerably reduced. A method for selecting the best terms, based on their inverse document frequency and term discrimination value, is also presented.

Many websites are in general poorly defined and its users are not able to find the information they need. That is the reason why many papers are addressed to propose techniques able to find the right information for a user. Most of these techniques focus on finding the required information in the whole Internet. Many times the owner of the website gives incomplete/imprecise information with low level of usefulness for the user. The re-structuring of the information is many times enough for detecting lacks of information, inconsistencies and imprecisions. However this work is normally very difficult without losing performances of the website. The authors have developed a novel application to exploit existing information in a website in a more profitable way restructuring the information without the intervention of the content provider. This paper describes the authors' experience during their participation in the European Commission ESPRIT 29158 FLEX Project.