Agriculture Knowledge Service System Based On Knowledge Graph

Agriculture knowledge service is the hots-pot problem in Artificial Intelligence filed and agricultural informatization filed.Agricultural knowledge question-answering is the one of the most important means for agriculture knowledge service.Traditional question-answering search answers according to keywords or shallow semantic parsing in lots of Web documents.Its answers are redundancy and can not satisfy users' requests.The sense of user experience is poor.Therefore,traditional methods actually did not understand users' intention.Based on this circumstance,the paper starts with agriculture knowledge graph construction and implements research of the related theories and technologies of agricultural knowledge question-answering.The following are main contents of the paper.(1)Construct Agriculture pests knowledge graph.According to structure of knowledge graph,the paper gradually constructs knowledge graph.Ontology is start generated by data classification standard.And then entity layer is generated by extend on ontology layer.Knowledge graph is formed.Besides,in order to visualize graph and increase expansibility,the paper studies knowledge graph storage with graph database Neo4 j.Agriculture pests knowledge graph provides supporting of knowledge base for knowledge question-answering.(2)Propose entity linking algorithm based on topic model and graph.Any knowledge graph can not cover all of entities in the real world.Entity linking is not only the crucial technology of knowledge graph extension,but also the key step to implement knowledge question-answering.Therefore,in view of considering the problem of ignoring combination of entity contexts and semantic correlation between entities,the paper links entity by simultaneously considering topic model and entity semantic graph and introducing entity similarity as node weight.Experimental results demonstrate that ELTMG has a better precision on entity linking.(3)Study knowledge question-answering based on knowledge representation and development agricultural pests knowledge question-answering system.In the method,agriculture pests knowledge graph is a basic knowledge base and the DBpedia is supplementary.The author explores the using of ELTMG to reduce the range of query and use word2 Vec and TransE model to train entity representation.In the paper,the method of knowledge question-answering is used to develop knowledge question-answering system.The system contains the module of knowledge question-answering and display module of knowledge graph.It aims at providing precision answering to users' question.