Research On Academic Quantitative Evaluation Based On Fine-Grained Knowledge Graph

Academic evaluation is a very important value evaluation work in scientific research activities.It is a key way to measure the academic quality and innovation of scientific research subjects.On the one hand,fair and objective evaluation of academic achievements can protect the vital interests of each scientific research scholar,give full influence to scholars with outstanding contributions,and promote the continuous development of high-end talents.On the other hand,it can provide academic guidance for researchers and improve the efficiency of researchers in analyzing the value of academic achievements and grasping the latest research trend.In addition,academic evaluation also affects the formulation of national scientific research strategy and the improvement of scientific and technological innovation ability at the macro level.It plays a very key role in promoting the healthy development of academia,ensuring academic integrity and rational allocation of academic resources.As one of the most important ways to display scientific research achievements,academic literature contains a lot of knowledge that is of guiding significance for followup research.At present,the qualitative evaluation method and quantitative evaluation method for academic literature are based on the title characteristics at the external level of the literature,such as cited quantity and influence factors,without fully considering the fine-grained knowledge entity characteristics at the internal text level of the literature,so they can not fully reflect the innovative value of the literature at the semantic level.In addition,the current knowledge graph used for academic evaluation are based on the characteristics of titles,such as citation network or co authorship network,and do not fully consider the correlation between fine-grained knowledge entities in literature.Therefore,how to analyze and mine the internal knowledge entity characteristics of literature from the fine-grained level,and on this basis,combined with the network relationship of academic knowledge graph,carry out academic quantitative evaluation on literature and scholars,is a difficult problem that has been concerned and discussed by the current academic circles.In view of the above problems,this paper takes the English literature in the field of pedagogy as the research goal,uses machine learning,natural language processing to design and construct the academic knowledge graph in the field of pedagogy from the finegrained knowledge entity level of the literature,and based on the constructed academic knowledge graph,combined with the knowledge entity characteristics of the literature and the academic network relationship of the knowledge graph,The design and application of academic quantitative evaluation algorithm for literature and scholars provide a new research perspective for the field of academic quantitative evaluation.The main research work of this paper is as follows:1.Design of fine-grained academic knowledge graph in the field of Pedagogy:This paper puts forward a design scheme of fine-grained academic knowledge graph in the field of pedagogy.Firstly,the program selects the field of pedagogy as the main research object of the paper,obtains the academic literature data set in this field,analyzes the vocabulary level characteristics of the literature in the initial data set through text cleaning,word frequency statistics,weight calculation,word vector generation,clustering and other methods,and abstracts six explicit knowledge entity categories contained in the literature,Then,combined with the needs of professional scholars in this field,the three categories of invisible knowledge entities are supplemented to obtain the category system tree of literature knowledge entities in the field of pedagogy.Finally,the overall framework design of fine-grained academic knowledge graph in the field of pedagogy is completed by designing the data types and relationship types of scholars,literature and knowledge entities.2.Construction of fine-grained academic knowledge graph in the field of Pedagogy:This paper puts forward a construction scheme of fine-grained academic knowledge graph in the field of pedagogy.Firstly,based on the abstract knowledge entity categories,a high-quality knowledge entity extraction task data set is constructed by using the manual annotation method.Then,based on the data set,combined with the points of knowledge entities in pedagogical literature,a knowledge entity extraction model based on dual task fusion strategy is proposed.The model can jointly model the knowledge entity extraction task through sequence annotation task and reading comprehension task,and process the results of the sub model based on two-level fusion strategy,so as to solve the problems of long-distance dependence and entity nesting in the process of knowledge entity extraction to a certain extent.Finally,based on the designed knowledge entity extraction model,batch knowledge entities are extracted from unlabeled literature,and the extracted knowledge entities,scholars and literature data and their relationships are imported into the graph database to complete the construction of fine-grained academic knowledge graph in the field of pedagogy.3.Academic quantitative evaluation algorithm based on fine-grained knowledge graph:Based on the constructed fine-grained academic knowledge graph,the academic quantitative evaluation algorithm is designed and applied for literature and scholars.Firstly,aiming at the literature,a literature innovation evaluation algorithm based on literature similarity calculation and hidden Markov model is proposed.The algorithm can calculate and evaluate the literature innovation value at the fine-grained level of literature content.Then,for scholars,a clustering algorithm of scholars' research tendency based on GN graph clustering is proposed.The algorithm can cluster scholars' research tendency by combining the internal knowledge entity feature relationship of literature and the network relationship of academic knowledge graph.Finally,the effectiveness of the above two algorithms is verified by experimental analysis.The experiments show that the two academic quantitative evaluation algorithms proposed in this paper have certain effects.