Reserch And Implementation On Semantic Processing System Based On Rule Matching

With the improvement of technology and quality of life,we are eager to solve problems through smarter and more convenient methods because of our increasing requirements for intelligence.In daily life,people obtain the required information through graphical interface which requires multiple cumbersome operations on the interaction,and the accuracy of the result cannot be guaranteed.The more convenient way of the intelligent era is to judge the user's intention and complete the user's needs only through one interaction of the user.In view of the above problem,this paper allows users to interact through establishing semantic framework model of voice and text,which can analyzed and output as structured semantics,and then complete the processing of the user's intention.This paper completes the analysis of spoken semantics and the execution of semantics.In order to convert semantics into computer-processable structured data,I put forward that the semantics are analyzed into three elements: domain,intent and word slot through the self-built semantic framework model.In the process of semantic parsing,a rule matching network is generated based on a finite state automaton,which hold a high accuracy rate.In order to reduce the complexity of rule matching,a rule grouping concept is proposed,and a rule configuration system is implemented to simplify the rule configuration process.To make up for the shortcomings of rule matching,a named entity recognition system is designed based on conditional random field algorithm to improve the recognition rate of unregistered words.Based on the semantic analysis,the structured semantic data is used as an instruction to construct a set of independent semantic execution systems to complete the semantic intent distribution and processing operations.Finally,each functional module is deployed and tested for functional effects,and compared with the semantic analysis method based on statistical model,with better precision and recall rate,and the accuracy of verification semantic business processing is up to 90%.