Research On Restricted Natural Language Query Interface Based On Dependency Syntax Analysis

Natural language is the most important information carrier in human society.Most of the data stored and exchanged in the real world is described by natural language.At the same time,it is also an important medium for human-machine interaction.The natural language query interface to database allows people to use natural language for data retrieval without any prior knowledge of the formal query language,database model or underlying data,and is therefore important for storing and retrieving information.The ideal natural language interface should support any query that the user makes,and pay the lowest cost to get accurate information.Due to the limitation of natural language processing technology,in practice,the scope of support for query statements is generally limited to a subset of natural language constrained by vocabulary,sentence,etc.,and a restricted natural language query interface is implemented within the scope.The research work in this paper mainly includes the following aspects:(1)A general database natural language interface model is designed to separate the knowledge base construction from the source code in programming.According to the abstract model of the general interface,the transformation from natural language to formal database language is decomposed into a two-layer mapping which from natural language to a conceptual model,and then to a structured language.It is divided into three modules: lexical analysis,dependency analysis and structured query language generation.System scalability is achieved by building a dedicated knowledge base.(2)This paper adopts the N-shortest path segmentation method combining statistics and dictionary,sets the path weight according to the priority of knowledge base,and marks the semantic label of the relation data for each word in the sentence.At the same time,a semantic annotation method based on semantic tags is proposed to realize the semantic coverage of query.The structural information of the query is obtained by the dependency analysis,and the semantic ambiguity of the data is eliminated by combining the SQL language features.(3)This paper takes the relational data semantic model as the conceptual model,and uses the set and linked list as the intermediate data structure,as well as analyzes and summarizes the syntactic features of common query sentences,and combines the rules to realize the recognition of target phrases and conditional phrases.A targeted transformation strategy is proposed for different objectives and types of conditions.The prototype system is applied to student information management,and experiments show that the model has certain availability.