Research On Chinese Phrase Structure Ambiguities Based On Semantic Analysis And Its Implementation

Ambiguity question is difficult in natural language processing, and puzzles understanding ability of the computer to the natural language. In this thesis, some research have been done about phrase structure ambiguity. Proposed the method that Knowledge Graph expresses the semantic knowledge, and applied the improved semantic knowledge resource-Hownet to the system of the phrase structure disambiguation.Multiplicity and complexity of Chinese ambiguity are still a difficult aspect on natural language understanding, especially the phrase structure ambiguity;it has already become biggest factors on the syntax analysis efficiency and the rate of accuracy. In this thesis, after summarised phrase structure ambiguity type;detailed analyzed and compared which the former phrase structure disambiguation, proposed the method that phrase structure disambiguation based on semantic. After different semantic theories and different semantic knowledge representations are analyzed and compared, we proposed the method that Knowledge Graph expresses semantic knowledge. Word graphs, phrase graphs are concluded in Knowledge Graph, Word graphs are automatic generate according to Hownet, phrase graphs are combined by Word graphs.In this thesis, basic semantic computations have applied to phrase structure disambiguation. After analyzed computed principle based semantic dictionary, proposed computed method that word similarity based knowledge graph, in this computed process, considering Knowledge Graph's characteristic and hierarchy of Hownet dictionary. Not only computed the first sememe similarity, but also other sememe. Improved the result accuracy and reasonable by this way.In this thesis, proposed method of phrase structure disambiguation, our work has gone deep into the semantic level of phrase structure disambiguation from the syntactic level. Applied powerful semantic knowledge -Hownet and new method of Knowledge Graph, implemented Chinese Phrase Structure disambiguation Based onSemantic. The parsing result from the testing corpus is satisfying.In the end, the future research work on phrase structure disambiguation is pointed out.