My research is a part of Data Minig field and Semantic Web field. More precisely, I am working in the Association Rule Mining field dealing with the extraction of implicative tendencies between items from a database. Unfortunately, the usefulness of association rules is strongly limited by the huge amount of delivered rules making very difficult for a decision maker to manually outline the interesting rules. Thus, it is crucial to help the decision maker with an efficient reduction of the number of rules. During my PhD I have been interested in the phase of post-processing of the rules, with a main goal – to integrate user beliefs and knowledge in the post-processing task in order to extract only those rules interesting the user.
At this end, I have proposed a new approach, presented in my recent publications, integrating 2 types of user knowledge representation:
1. A general view over user knowledge in database domain is described using ontologies. I have proposed ontologies as user knowledge representation because they are the most complex structure for user knowledge representation proposed in the literature. Comparing to taxonomies, ontologies offer a more complex knowledge representation model by extending the only is-a relation present in a taxonomy with the set R of relations. In addition, the axioms bring important improvements permitting restriction concept definition starting from existing information in the ontology.
2. I have proposed a new formalism - Rule Schemas - in order to represent user expectations concerning extracted rules. The base of Rule Schema formalism is the specification language for user knowledge introduced by Liu et al. (1999). The model proposed by Liu et al. (1999) is described using elements from an item taxonomy allowing an is-a organization of database attributes. Using the items in a taxonomy as elements of the specification language has many advantages: the representation is more general and the schemas more abstract. However, a taxonomy of items might not be enough. The user might want to use concepts that are more complex than generalized concepts and that result from relationships other than the is-a relation. This is why we have considered that the use of ontologies would be more appropriate. An ontology includes the features of taxonomies but adds more representation power. In a taxonomy, the means for subject description consists essentially of one relationship: the subsumption relationship used to build the hierarchy. The set of items is open, but the language used to describe them is closed by using a single relationship (the subsumption). Thus, a taxonomy is simply a hierarchical categorization or classification of items in a domain. On the contrary, an ontology is a specification of several characteristics of a domain, defined using an open vocabulary.
Furthermore, these propositions are integrated in an iteractive framework.