Research On Semantic Matching Model Based On Concept Lattice

In the current age of "information explosion",how to retrieve the needed information conveniently, accurately and quickly has become the focus issue. However, the essence of today's information retrieval is the "keyword matching operation".Therefore, how to establish the semantic matching model to meet the personalize information needs etc. is the direction of search engine in the future.Concept lattice is a concept hierarchy structure which built upon the binary relation between the data set objects and attributes; it vividly embodies the simple concept of generalization and specialization among relations, while its Hasse diagram has achieved data visualization.The central task of this paper is to apply the mathematical description of the concept lattice to concept in information retrieval, particularly on how to build the concept lattice based on the formal context for the semantic matching model.This paper tries to combine the formal context to define semantic matching model, and define the document as the object set form of formal context, while attribute set is a subset which can represent the key word of the document characteristics. Thus to extract the conceptual hierarchy connection between the document set and key word, between documents and document, and use the concept lattice to reflect the relationship between formal concepts. The matching model can navigate the users through the view on the concept lattice. It is natural for us to use the Godin algorithm because the attribute set would change accordingly when the document based on the object centralized is added dynamically. The object set and the attribute set of the formal context decide the structure of the concept lattice, which will directly affect the user's browse efficiency and the recall ratio and precision ratio of the semantic matching model. Therefore, the most critical step is to establish formal context, namely, the matching model. Attribute set selection determine the performance of the matching model.Finally, an example analysis is done to prove the feasibility of concept lattice-based semantic structure matching model with the formal context through the attribute extraction. The superiority of the model lies in that it can reflect potential of clustering relations between documents for the organization of data source. With the combination of attribute reduction in concept lattice, the model provides users with a semantic matching model based on concept lattice of theoretical value.