Chinese Semantic Analysis System Based On The Concept Map Study And Realization

Semantic analysis plays an important role in many areas of Natural Language Processing, and now it has become a hotspot and nodus. It is supported by the knowledge of syntax analysis and knowledge representation. At the same time, it is need to take into account of the theories in the domain of linguistics, psychology, philosophy and so on. It is helpful for information retrieval, machine learning, text generation, question answer and etc.The main work in this dissertation is to study the semantic analysis in Chinese. A system about machine learning and question answer has been built at the same time. The main work and innovative results of the dissertation is organized as follows.Firstly, introduce the research status of the overseas and the domestic. Also explain the object and the goal of the work. The dissertation mainly about the analysis of Chinese real text sense, and the system try to imitate the process of human how to get the knowledge and how to use it.Secondly, do some research in the domain of knowledge representation and reasoning. Expand the theory of Conceptual graphs according to the features of Chinese. Base on this structure we can make semantic valuation. Furthermore, summarize a set of rules to translate the grammar relations into semantic relations.Thirdly, design and implement the semantic analysis system. A semantic analysis model has been given in this section. There are six modules which are applied in the system: pretreatment, translation, display and modify, question generation, calculation and output. In the module of pretreatment we use the HIT IR Lab resources to participle and line out the Chinese text. Translation is the core of the whole system. A set of rules are summarized to be used in translating 24 grammar relations into 49 semantic relations. The results can be used in generating Conceptual graphs. In the display and modify module we can correct the mistakes of the Conceptual graphs. Question generation is to translate the questions into the same structure on which we can search the answers. Output is to translate the result Conceptual graphs into natural language. We use text and voice to output the result.Lastly, give an evaluation of our semantic analysis system. The experiment results have been proved that the model is effective in semantic analysis and knowledge learning. It make a better recall than other systems.