Information Retrieval Research And Implementation Based On The Concept Semantic Similarity Computation Model

Currently, there are various defects in using the traditional information retrieval based on the simple keyword matching. The main reason is that the computer can not understand the words of retrieval requests and the user’s retrieval intent either. In order to make up the defects, it must use semantic retrieval to understand the semantics of information. And it has become a very challenging research topic now. During the realization of semantic retrieval process, the concept is the basic unit of information representation. Directly, the semantic retrieval performance is affected by the accuracy of semantic similarity between concepts. And, the concept semantic similarity computation is the key step of extending the concept semantic and quantifying the similarity between the query and the resulting documents. For the reason that, a superior computation model based on the concept semantic similarity is important and has great theoretical significance to semantic retrieval.To the basis of e-commerce domain ontology, the thesis proposes an improved computation model based on the concept of semantic similarity. The model integrates the two aspects of distance-based and content-based. Consequently, it can quantify the concept semantic similarity in the ontology comprehensively and accurately. Accordingly, it extends the concept semantic of keywords set and calculates the similarity between the query and the resulting documents. Finally, the retrieval algorithm of this thesis has been generated. In this thesis, it has achieved an intelligent retrieval system about e-commerce information with using this retrieval algorithm.The experiment shows that the proposed retrieval algorithm of this thesis has a better retrieval performance, comparing to the retrieval method based on Lucene.