Research On Key Technologies Of Campus Intelligent Question Answering System Based On Domain Knowledge Base

With the rapid development of the Internet,the speed of production and dissemination of information has increased dramatically.How to improve the efficiency of information acquisition has become a focus of widespread concern.At the same time,the emergence of an intelligent question answering system with high efficiency and convenience has greatly solved this problem.In the campus area,various information and services were distributed under different departmental portals in traditional information service.The information is messy and diverse and the acquisition method is inefficient.This approach has not satisfied the efficient demand of teachers and students.Therefore,this paper will study and implement an easy-to-use and efficient campus intelligence QA system as a unified entrance for campus information acquisition.It can not only provide accurate question and answer information,but also provide intelligent campus information services(inquiry results,classroom booking,etc.).This paper mainly studies the key technologies of campus intelligent QA system from the aspects of knowledge base design,question analysis and answer query,combined with domestic and foreign research results.The primary study in the paper are described as follow:1)Analyze the characteristics of the existing four types of intelligent QA systems,combining the needs of users in the campus area,a domain knowledge base model composed of hierarchical multi-label knowledge bases and domain knowledge graph was proposed.It solved the hard problems including multiple problem-solving methods,lack of question-sentence information,different granularity levels of questions,artificially constructed knowledge-base costs,and high demands on the efficiency and accuracy of QA in the campus intelligent QA system.2)The word segmentation based on domain dictionaries is applied to the question to improve the accuracy.A SVM classification algorithm is proposed that combines domain heads and parts of the question center headings and keywords as question character classification features.Its practicality was proved through experiments.It provides support for subsequent generation of answer search strategies based on label matching.3)Describe standards and specifications for service APIs,provided standard registration templates,enabled service API calls to be integrated into hierarchical multiple-tag knowledge bases,and automatically generated calls to corresponding service APIs based on the understanding of user issues.Intelligent QA form of campus information service(inquiry results,booking classrooms,etc.)4)A Semantic Dependency Analysis Method(R-SDP)based on reduced rules was proposed in this paper to obtains question label sets,and the corresponding label matching algorithm was proposed to query the answers,and realized the intelligent interaction.5)Based on the research of the related technologies,uses the existing development tools and framework to design and implement the WisQA prototype system.