A Refinement Operator Based Method for Semantic Grouping of Conjunctive Query Results

The methods proposed for aggregating results of structured queries are typically grounded on syntactic approaches. This may be inconvenient for an exploratory data retrieval, with often overwhelming number of the returned answers, requiring their further analysis and categorization. For example, if the values instantiating a grouping criterion are all different, a separate group for each answer would be created, providing no added value. In our recent work, we proposed a new approach, coined semantic grouping, where the results of conjunctive queries were grouped based on the semantics of knowledge bases (ontologies) of reference. Specifically, a user defined grouping criterion was expressed as a concept from a given ontology, and results grouped based on the concept subsumption hierarchy. In this work, we propose a novel method for the task of semantic grouping, that is based on an application of a concept refinement operator. This novel method is able to deal with some cases not handled by the initially proposed one, where, for example, a grouping criterion is a primitive concept thus not allowing for further semantic grouping of the results. In such a way, we achieve a solution able to deal with both problems: of too large and of too small number of groups.