Research On Word Sense Recognition Based On Statistical And Knowledge And Its Application In Mechanical Product Design

Intelligent is the development direction of the mechanical product design, the design of the intelligent product is the key to the "man machine interface". There are not many words in natural language, but the frequency is very high. Therefore, the word sense disambiguation of the meaning of a word is directly influenced by the deep understanding of the following. This paper will be based on the combination of statistical and knowledge of the meaning of a word sense recognition application to mechanical product design, through the right understanding of customer needs to achieve product design intelligence.Firstly, this paper describes the overview of natural language understanding and the current situation and development trend of the word sense disambiguation. Key of polysemous words caused by ambiguity analysis, for indicating the completion of ambiguity caused by temporal word polysemy, showing complete temporal ambiguity; for noun or adjective modifiers of polysemous words caused by the ambiguity and show the word is error cut apart caused by unable to determine polysemy is the right meaning of the word; for based on statistical methods to ambiguity caused by the disambiguation of polysemous words showed polysemy is the correct meaning cannot be determined. Then, based on the statistical method, it is generalized to complete the process of semantic ambiguity of words and phrases, and can not be determined by statistical method. Therefore, this paper constructs a semantic recognition model based on the combination of statistical and knowledge, which mainly includes the preprocessing module, the context acquisition module, the disambiguation module and the corpus management module. Among them, the function of disambiguation module is emphasized, which is to determine the correct meaning of the word. And polysemy of words recognition, is inseparable from the disambiguation corpus, the disambiguation corpus is in How Net based on the mutual information, the concept of connotation and denotation and event operation by constructing the combination and corpus management module is responsible for the maintenance of corpus and updated in a timely manner.Finally, this paper will build the model of the semantic recognition based on the combination of statistics and knowledge to the mechanical product design, to help enterprises or designers to analyze customer needs and verify the feasibility of the model.