Restricted Domain Question Answering Natural Language Database Query

Restricted Domain Question-Answering Natural Language Database Query(NLDBQ) is a means that the user in a particular area who want to have a database access method gives a question to computer by using natural language and the computer give this questions'answer. This means can be return to the user the accurately answers, it is a very effective question-answering implementation model, have broad application prospects. This article aim at the restricted domain natural language database query systems'characteristics, study the question-answering database query which with the field of tourism hotels, attractions, places, people and culture categories as the core content, mainly research NLDBQ questions'shallow semantic analysis, and identify the domain questions'semantic chunk, take this domain questions'semantic chunk as the middle language in order to complete the questions convert to the SQL. This article is designed and implemented the above method based on Restricted Domain Question-Answering NLDBQ prototype system. Specifically, this paper's main studies are as follows:Firstly, this article analyzed the natural language database query systems'important and difficult point, introduced the questions pretreated methods, and used the HIT LTP platform to analyze questions'syntactic structure, extracted the questions'main struts, and combined restricted domain questions characteristics, questions pretreatments and so on, completed a domain question rules-based classify method, this method combined questions main structure with domain question characteristics, lay the foundation for identify the domain question semantic chunk.Secondly, this article gives the domain question chunk definition and identifies method in the process of restricted domain question-answering NLDBQ. This method combined the characteristics of the restricted domain question-answer NLDBQ, definite four domain semantic chunk as follows:the domain question entity semantic chunk, domain question target semantic chunk, domain question condition semantic chunk and domain question focus semantic chunk, combined domain question analysis, identified the contents of the four chunk, take these contents as the middle language in process of NLDBQ, lay the foundation for the generation of SQL. Thirdly, this article combined the results of domain question pretreatment, question classify, question named entity recognition, question semantic chunk and so on, used the forms all domain question characteristics and domain question chunk analysis results in the database, to obtained the SQL statement for the database query, to give a method of domain question convert to SQLFinally, this article takes the field of tourism as the support area of Yunnan Province, realized restricted domain question-answering natural language database query prototype system. The system is mainly include the query as follows:based on single entity and single question focus query, single entity and many question focus query, many entity and single question focus query, many entity and many question focus query. Experimental results show that the method based on domain question shallow semantic analysis and take domain question semantic chunk as middle language has a significant effect to restricted domain question-answer natural language database query system.