Improved SME Modle Based On Multiple-valued Lexical Semantics Similarity

Structure Mapping Engine SME is an analogy inference calculation model which is proposed based on structural Mapping Theory SMT. It is mainly known from the knowledge that between the old knowledge and new knowledge to find a most similar one, use the similarity between the old and new knowledge structure, and establishing the connection between the old and new knowledge. The learners can learn new knowledge by use the known knowledge and the connection between the new knowledge. There are two problems in SME analogy inference calculation model, the first is that the new knowledge structure and the knowledge structure the need of human given, it makes generate knowledge structure process trouble, inaccurate; the second is to calculate the semantic similarity of words two values 0 or 1 ,it makes the analogy of the last generation can not determine the use cases, can not even produce an analog use cases.In view of the above two problems, this paper mainly based on research the relational data sources, and it will based on "Knowledge Network" in the form of semantic similarity calculation and the background added to the analogy with the structure model, with a formal context that the structure of knowledge, and generate improved model. The main research is done in this paper can be classed as follows:Firstly, The semantic similarity calculation method which based on " Knowledge Network t" is applied in analogical reasoning calculation model. The traditional model of analogical reasoning to calculate the value of lexical semantic similarity is 0 or 1, it makes the analogy to identify use cases use case is uncertain or can not find the analogy. According to this problem, the calculation method of semantic similarity which based on the " Knowledge Network " will be applied to calculate the improved model, the concept similarity value is a real number between 0 and 1, and the above defects is solved.Secondly, The form background isomorphism is applied in analogical reasoning calculation model. The traditional analogy inference model need of human knowledge structure and some new known knowledge structure, this makes the process of structural produce inaccurate, complex shortcomings. This article will form background as field knowledge structures, from the generation in the source semi-automatic relational, make form of knowledge structure of the background generated more convenient and accurate. Solve the defects of the above.On this basis, this paper carries on the improvement of SME model and produce improved model. Through questionnaires way, verified the model can provide an analogy cases generated than traditional model of analogy cases can help learners to learn new knowledge.