Research On Construction And Application Of Knowledge Graph Of Minority Festivals

China is a unified multi-ethnic country,and the long river of historical civilization has bred colorful national festivals.Each minority festival has its own unique cultural value,bearing the spirit and emotion of each nationality,and exploring appropriate digital methods to protect and spread excellent minority festival culture is essential.The knowledge graph is a structured semantic knowledge base,used to describe the concepts and their relationships in the real world in the form of symbols.At present,many fields have conducted in-depth research on the construction and application of knowledge graphs,but the application of knowledge graphs to the construction of minority culture,especially in the field of minority festival culture,is still at the preliminary exploration stage.Therefore,this article focuses on the characteristics of the field of minority festivals and the task of spreading minority culture,and studies the construction and application of the minority festival knowledge map,mainly including the following four aspects:(1)Research on the extraction method of minority festival information for unstructured text.Aiming at the task of extracting triplets in unstructured text,a pipelined triplet extraction model based on ALBERT pre-trained language model is proposed to optimize the traditional relationship extraction model.Increase the speed of prediction while transmitting.The model is verified on the self-built data set to complete the knowledge extraction task well.(2)Research on the construction of minority festival knowledge graph.First,obtain the original knowledge data from different platforms,propose an improved ontology construction process and construct the initial ethnic festival domain ontology;then use multi-source heterogeneous data to obtain ethnic festival triples based on ontology constraints and use targeted knowledge extraction methods,and A pipelinebased knowledge fusion scheme is proposed to remove redundant knowledge;finally,the triples are saved using Neo4 j graph database to complete the initial construction of the minority festival knowledge graph.(3)The construction of a question-and-answer model of minority festival knowledge based on the knowledge graph.This paper proposes a knowledge graph question and answer model that combines deep learning and template matching.Through semantic analysis such as entity recognition,intention recognition,and type judgment,natural language questions are converted into Cypher query sentences,and a question test set is constructed to verify the correct conversion rate.At the same time,the details of the question and answer of the model are also optimized to improve the human-computer interaction experience of the model.(4)Design and development of minority festival cultural resources knowledge service platform.Based on the above work,design and develop a knowledge service platform for minority festival cultural resources based on knowledge graph.This platform can not only provide technical support for information extraction such as named entity identification,relationship identification,and triplet extraction,but also provide knowledge graph visualization retrieval and minority festival knowledge question and answer system,online management of knowledge graph and other applications.