Research And Application Of Key Technologies In The Construction Of Cultural Knowledge Graph

With the development of social economy,people’s living standards are constantly improving,people’s ideology is becoming stronger and stronger,and their pursuit of the spiritual world and attention to culture,especially traditional culture,are gradually increasing.With the help of knowledge graph,a graph-based data structure,the relationship between entities in traditional culture can be more intuitively displayed in the way of nodes and edges.Traditional culture has a large number of knowledges that has not been fully excavated hidden in unstructured text.This paper takes traditional culture as the theme,and takes the key technologies in constructing the knowledge graph in the cultural field as the research point.This paper focuses on the named entity recognition algorithm and relationship extraction algorithm for extracting information from unstructured text and constructing knowledge graph.The algorithm is applied to construct the service platform of knowledge graph in cultural field,which plays a guiding role in the construction of knowledge graph in related fields.The main work of this paper includes:(1)Design and implementation of named entity recognition method in cultural field.In order to extract keywords from unstructured text as the data basis of knowledge graph construction.This paper analyzes the characteristics of ontology in the field of traditional culture,and designs a general ontology design method for the field of traditional culture.Taking traditional architecture as an example,the feasibility and expansibility of ontology construction method are tested.Through manual data collection and annotation,the named entity recognition data set of traditional buildings in Beijing is constructed.With the help of the pre training model,a named entity recognition algorithm based on BERT-BiLSTM-CRF is proposed.Compared with other existing models,it is proved that this model can effectively improve the recognition effect.(2)Design and implementation of relationship extraction method in cultural field.In order to further improve the automation of constructing knowledge graph,a relationship extraction model based on BERT-BiLSTM-Att is designed.Experiments show the effectiveness of the relationship extraction algorithm.(3)Design and implementation of knowledge graph service platform for cultural field.The system allows users to interactively query and browse knowledge related to culture.It also allows users to ask questions about virtual images with the help of natural language,and continuously answer questions in areas of interest to users in the form of dialogue,so as to protect and spread excellent traditional culture.