Classification Of Academic Resources Based On Graph Attention And Construction Of Scholar Graph

In recent years,academic research has been supported by all sides of society,and a large number of research results are emerging.It is particularly important to classify academic resources quickly and effectively.Academic resources are different from general information resources.On the one hand,there is a wide range of information sources,and information is freely released.On the other hand,there are many and diverse types of academic resources.The knowledge graph can unify the structure of different data sources,effectively integrate resources,organize and manage information.When constructing the graph,this paper not only considers the semantic information of scientific research achievements,but also mines the relationship information in the graph.From the perspective of classification,this paper proposes a classification model of academic resources based on graph attention and constructs a scholar graph,which improves the accuracy of resource classification and the practicability of scholar graph.The specific research work of this paper mainly includes the following aspects:Firstly,we propose an Association Content Graph Attention Network(ACGAT)based on association features and content attributes to classify academic resources.Semantic association and academic association are introduced into the model.In the dimension of resource association characteristics,on the one hand,the model mines the association commonality among academic resource nodes to improve the aggregation of the network by reducing the existence of isolated nodes in the existing graph attention network.On the other hand,the model calculates the influence of the node,which enhances the positive effect of the node on the network.In the dimension of content attribute,the model extracts the text semantics of academic resources,mines the semantic similarity of nodes,which enriches the content expression of resource nodes.Finally,the ACGAT model uses the attention mechanism to aggregate and update the features of academic resources to improve the accuracy of classification.Secondly,based on the classification results of academic resources,we construct the scholar graph and propose the formal definition of the scholar graph.The nodes in the graph represent scholar entities,and the connections between nodes are the relations between scholars.The construction of scholars’ portraits mainly includes three dimensions: the basic attribute dimension of scholars,the research field dimension and the cooperation preference dimension.The basic attribute information of scholars can be obtained by crawler technology.In this paper,the LDA model and statistics are used to obtain the dynamic change trend of scholars’ research interests in years.For the relationship between entities,by analyzing the citation relationship of the paper and the cooperative relationship,found the direct reference relationship,co-citation relations and cooperative relationship between entities.We propose the Multi-signature Cooperation Contribution Model based on H-value(MCCM＿H)to measure the collaborative contribution of scholars,which is helpful to excavate the cooperative preference information among scholars and construct a rich scholar graph.Thirdly,in order to verify the feasibility of the proposed academic resource classification based on graph attention and the construction of scholar graph,we make the experimental analysis in the self-built dataset SIG and the public dataset including Cora and Citeseer.Compared with other models,the experimental results verify the accuracy of the proposed graph attention model based on association features and content attributes in academic resource classification.Then we analyze the applicability and rationality of the scholar graph based on resource classification.