Knowledge Base Construction For Car Evaluation Based On Fuzzy Concept Lattice

A large number of reviews is produced with the development of web 2.0.These comments guide users choose products,and help producer get feedback.But there is difficulty with how to get useful information from reviews.So extracting knowledge and constructing knowledge base for product evaluation is significant.This thesis has a complementary fusion in theory across formal concept analysis,fuzzy logic and association rules,then defines the relevant concept of car evaluation knowledge,finally puts forward the construction method and application mechanism of knowledge base for car evaluation based on fuzzy concept lattice.The construction of knowledge base for car evaluation from reviews has potential application value in the network text opinion mining,development of product evaluation and recommendation system.In this paper,although the experiment is conducted on car evaluation text,the method is universal,has a good reference value to related research.The main research contents is as follows:(1)Knowledge definition for car evaluation based on formal concept analysisIn this paper,we define the basic concepts and knowledge about the knowledge base for car evaluation.We combine the theory of formal concept analysis and fuzzy logic which are describe the uncertain information,put forward the fuzzy concept lattice suitable for car evaluation.We use fuzzy association rules lattice redefine the parameters in the extraction process of fuzzy Association rules.It not only enriches the theory of formal concept analysis,but also provides the theoretical basis for the later construction of knowledge base for car evaluation.(2)Knowledge base construction for car evaluation based on Fuzzy Concept LatticeReviews contains abundant evaluation knowledge.To extract and organize the knowledge effectively,a novel construction method of knowledge base for car evaluation using fuzzy concept lattice is presented.First,according to ontology,evaluation collocations for building 5-tupel arrays to generate fuzzy formal context are extracted from the comment text.Then the incremental algorithm was adopted to construct concept lattice.On the basis,fuzzy concept lattice as a core of knowledge base is formed by adding the membership values of characteristics.A knowledge base is constructed with 37646 comments of 374 cars price range between 50000 and 200000 from www.auto.sina.com.cn by designing the threshold value T.Finally,using fuzzy concept and lattice structure,knowledge can be represented and organized objectively and efficiently by the analysis of the data in the knowledge base.(3)Knowledge construction and application for car evaluation based on Fuzzy Association RulesKnowledge and hierarchical relationships are described by concept and the lattice construction in car evaluation knowledge base based on fuzzy concept lattice.This paper utilizes fuzzy association rule lattice theory to improve the knowledge base,redefine the parameters in the extraction process of fuzzy Association rules,then extracts fuzzy association rules,finally establishes the association knowledge.The association knowledge can improve the knowledge base.According to the fuzzy association rules of constructing retrieve keywords,the default information is deduced in car evaluation knowledge base.