Research And Implementation Of Entity Potential Relationship Knowledge Graph Based On AttCNN Model

With the development of the information age,data presents an explosive growth.The realization of data visualization helping to find the trend of data from the complex information,is a global view of the data,and can realize a more accurate grasp of the data.Moreover,the large amount of network data is easy to obtain,which can better meet the demand of knowledge graph for data sources.To make the best use of data sources,extract information and visualize it to the maximum extent,knowledge graph is the first choice.The introduction of knowledge graph plays an important role in the construction of discipline research platform.Most of the research on knowledge graph focuses on relation extraction and visualization technology of knowledge graph.The research still focuses on the existence relationship,that is,the association relationship between entities,while the research on the unassociation relationship between entities is a little insufficient.This paper starts from this aspect to study the entity relationship of the knowledge graph.Based on the accumulated theoretical research and practice of knowledge graph,this paper selects relevant technologies to construct a complete technical scheme of knowledge graph,and studies the potential association between entities in the knowledge graph.Firstly,this paper uses rich association to extract entity relations,that is,in the context,multiple entities are extracted,and the eigenvalues of entity relations are extracted through convolutional neural network and attention mechanism,and then matched and fused.Then,according to the obtained triples,the unassociated entity relations are extended and the blank set of entity relations is given.Finally,the potential relationship between entities is studied through the blank set of entity relations.In this paper,the Neo4 j graph database is selected as the information storage tool,that is,the triplet relationship between entities extracted from the context sentence pattern is stored.In the system,the user can enter related questions or statements in the web page to carry out entity identification,and the system returns the identification results according to the user's questions.In this paper,literature data of relevant disciplines were selected for experiments on the CNKI.The experimental results show that the knowledge graph of potential relationship between entities designed and implemented in this paper based on the ATTCNN model has a good accuracy and efficiency in identifying entities,which can display more entity relationship information and enrich and improve the knowledge graph.