Semantic Search By Matching Conceptual Graphs

The tremendously rapid development of the Web has made it the largest information base that human beings have ever had. However,"information overload"is reducing the usability of it. When people turn to depend on search engines and information retrieval techniques to locate information they want, they tend to discover that the current methods based on keyword matching often fail to well meet their information needs. It is mainly because merely using keywords is not sufficient for clearly describing the semantics in user queries. Even by adopting query expansion or word sense disam-biguation, the issue still exists, which requests the emergence of"semantic search".Characterized as cross-disciplinary, semantic search has not yet got a well-acknowledged definition. However, the noticeable development in natural language processing, text mining, knowledge representation and in-ference, etc., together with the rising of the Semantic Web, has greatly driven the researches on the topic of"semantic search". The semantic search this paper tries to attack can be defined as a kind of search technique that per-forms query content matching based on explicitly and formally defined semantic information. Aimed at resource search by matching the content of text descriptions, we choose conceptual graphs as the semantic representa-tion in our approach, after we compared a variety of knowledge representa-tion languages that are commonly used in semantic search. Conceptual graphs are a graphic representation for logic with the full expressive power of first-order logic. Hence, the issue of semantic matching can be reduced to graph matching, without loss of the rigor in logic.The main concerns and contributions (innovations) of this paper can be...