Research On Domain-based Question Answering System Based On Ontology

After each year's college entrance examination,there will be a large number of candidates and parents to ask questions about the institutions and professional choices.However,most candidates and parents do not understand the various enrollment majors when filling out the college entrance examination volunteers.Blindly choosing the majors will not only increase the risk of candidates withdrawing from the exam,but also have a significant impact on the candidates' future career direction.Although candidates and parents can obtain information from the Internet by keyword search,this method is prone to generate a lot of messy and irrelevant information,and the retrieval efficiency is low.Therefore,this paper designs and implements a question and answer system for the enrollment consulting field of Yunnan University of Finance and Economics.It accepts users to input and return answers in natural language,which can effectively answer questions about candidates and parents about college majors.The ontology's ability to describe semantic information and reasoning ability can greatly improve the accuracy of the question and answer system.Therefore,this thesis uses ontology technology to integrate the relevant data of college enrollment majors and university functional departments,and constructs an ontology oriented to the enrollment consulting field of Yunnan University of Finance and Economics.Then,the natural language questions input by the user are preprocessed,including word segmentation,part-of-speech tagging,subject recognition,and question classification.In order to solve the problem of new word recognition in specific domain texts,based on the syntactic analysis,combined with the expression form of knowledge in ontology,and the recognition rules of proper nouns in the domain,this paper proposes a domain ontology and syntactic analysis.Combined with the hybrid word segmentation algorithm,the experimental results show that the hybrid word segmentation algorithm can better solve the problem of identifying new words in the field.The subject recognition model based on CRF algorithm is used to identify the organizational name entity of the university in the sentence,and the attribute predicate in the sentence is extracted by syntactic analysis technology to construct the SPARQL query triplet.The question classification uses the LSTM model-based problem classification algorithm to classify the questions input by the user,which is convenient for the program to call different SPARQL query templates.Then,using the constructed enrollment consulting domain ontology as the knowledge base of the question and answer system,using the dependency syntax analysis technique to extract the subject,predicate and object needed to construct the triad from the sentence,and convert it into SPARQL query triplet,and then combine The user dictionary and the domain ontology map the SPARQL query triples into ontology triples,and finally realize the answer retrieval based on the ontology knowledge base.Finally,using the web development framework Django,combined with the front-end development technology HTML+CSS,completed the web portal visual display of the enrollment consultation question and answer system.The question and answer system for the enrollment consulting field of Yunnan University of Finance and Economics designed and implemented in this paper can not only answer the questions of candidates and parents about the enrollment major of Yunnan University of Finance and Economics,but also inform candidates and parents about some important attributes of the functions of Yunnan University of Finance and Economics.Information,such as "How much is the duty phone of the Finance and Security Office?","How about the faculty of the School of Statistics and Mathematics of the University of Finance and Economics?","Where is Yunnan University of Finance and Economics?","Where is the School Affairs Office?",“Which college is the visual communication design profession?”,“How is the tuition fee of Yunnan University of Finance and Economics?”,“How much is the telephone number of the University of Finance and Research Hospital?”,etc.,the question and answer system can reduce Yunnan to a certain extent.The labor cost of the University of Finance and Economics Admissions Advisory Service has important practical research significance.