Research On Automatic Question-answering Of Relatives In Chinese Historical Figures

The search engine and the retrieval method constantly update the iteration in the demand for fast and accurate information acquisition.On the one hand,along with the vigorous development of the digital industry,has produced more and more digital resources.On the other hand,with the development of the automatic question-and-answer technology,users are no longer satisfied with the query of keyword matching,and start to search for a more concise and accurate retrieval engine.China Biographical Database is the tip of the iceberg of digital industry,how to make the digital product pour into the soul of the intelligent Q&A,making synthetic products better service for the user.This paper makes a preliminary research and exploration.In this paper,we study the method of the relation of inferential defect,realize the question and answer system,solve the problem that the retrieval engine does not support the natural language retrieval,and returns the result is not simple.The main work of this paper is as follows:First,research on character relationship reasoning.This paper studies the knowledge base of learning translation model,through the contrast test of the existing various translation model to find the right representation method of knowledge base,realize the relationship between the characters knowledge map reasoning,effectively solve the China Biographical Database characters in the kinship of the problem of the missing.After that,the comparison and analysis of the relationship between the relatives of the characters and the reasons for the possibility of reasoning errors were analyzed.Second,improve the translation model.Translation model is a method to transform the entity and relation of knowledge base into low dimensional vector.In this paper,on the basis of the existing models of translation,by improving the way of the negative samples of physical characteristics,a vector entities and relations in space distribution more reasonable,more expressive,in theory,this method can effectively improve the accuracy of relational reasoning and its feasibility in the China Biographical Database effectively.Third,design and implement Q&A system.This paper uses the improved translation model to design and implement the question and answer system of kinship in Chinese history.With China Biographical Database(CBDB)as the source of corpus,The system converts the kinship relation of Chinese historical characters stored in the access relational database into the knowledge based of the kinship relationship stored by the Neo4j graph database.This facilitates visual display and analysis of character relationships.This system uses natural language processing technology to generate Cypher query statements,query,reason and return questions in the graph database.