Query-Answering CG Knowledge Bases

Abstract

Conceptual graphs are a good choice for constructing and exploiting a knowledge base. In several of our projects (semantic portal for e-tourism, exploitation of digital object corpus, etc.), we have to query such bases. So it is natural to consider queries and bases as simple graphs and to compute the set of all projections from a query to a base. However, there is a problem of the return of this set of projections to the user. More generally, the main issue is about the definition of the notion of answers in an query-answering system made of knowledge bases formalized by graphs (Conceptual Graphs, RDF (Resource Description Framework) , Topic Maps, etc.). In this paper, we study several notions of answers and some of their characterizations. We distinguish between notions of answers by subgraphs of the base and answers by creation of result graphs. For the last type of answers, we define completeness, non-redundancy and minimality criteria of the answer sets and propose several notions of answers w.r.t these criteria.

Distinguishing Answers in Conceptual Graph Knowledge Bases

“…there is no answer that subsumes another one). This corresponds to our previous notion of redundancy defined in [2].…”

“…In ICCS'08 [2], we identified the semantic database context in which a set of answers has to be computed. For this case, there are two kinds of answer graphs: answers as subgraphs of the knowledge base graph that are used for browsing the KB or for applying rules; and answers as graphs, independent of the KB, corresponding to the classical use of a querying system 1 .…”

“…Indeed, the presence of unspecified entities (generic markers in CG or blank nodes in RDF) and also a type hierarchy leads us to consider that an answer which is more general than another as redundant. In [2], we studied this problem and proposed to return irredundant forms of the most specific answers to a query.…”

“…In [2], the completeness criterion (the knowledge expressed by each initial answer is expressed by one of the answers contained in the returned subset of answers) ensures that no knowledge is lost when a redundant answer is deleted. But this redundancy is based only on the comparability of answer subgraphs (that are semantically close as they are specializations of the query).…”

“…Indeed, two answers can appear redundant when they are not really redundant. In [2], we define the notion of redundancy based only on the comparability of the answer graphs. This approach was motivated by the following argument: as the returned set of answers is independent of the KB, the subset of the more specific irredundant answers, denoted R min , is sufficient to bring the entire range of answers.…”