Construction And Application Of The Knowledge Graph Of College Student Competency Based On Deep Learning

The cultivation of competency plays a vital role in improving students'quality and developing their career paths.With the popularization of higher education,people increasingly place their hopes on colleges.In this context,it is increasingly essential to provide competence-oriented knowledge services for students.As the main body of talent training,colleges and universities produce many teaching data in teaching activities,which plays a guiding role in cultivating student competency.Simultaneously,various job information published on the internet by enterprises has gradually become a vane of students'competency training.However,the description of professional vocabulary has a particular domain.Teaching data and network data are very complex.When learning competency-related knowledge,students usually need to spend lots of energy to retrieve and integrate information.Besides,it is not easy to embed this scattered knowledge into the process of competency cultivation.Therefore,integrating,managing,and applying knowledge in the multi-source heterogeneous data related to college student competency is a research problem of great significance.The development of knowledge graph provides an excellent solution for data integration.In this paper,the competency knowledge graph of college students is constructed by using related technologies,hoping to provide data support for intelligent applications related to student competency while facilitating students'learning by integrating relevant data.The main research contents of this paper are as follows:(1)By analyzing and processing the multi-source heterogeneous data,this paper defines the data schema of the college student competency knowledge graph,which includes seven kinds of entities and eight kinds of relationships.Among the entities,the three core entities defined for student competency are knowledge,skill,and dispositions.(2)For the task of entity recognition,this paper combines the BERT pre-training model with the BiLSTM in order to automatically extract the rich word-level features,grammatical structure features and semantic features in the input sequence,and complete the classification of competency-related entities.The BiLSTM-CRF model is used as comparative experiments for entity recognition in this paper.The results show that the precision,recall,and F₁ value of our model are improved by 0.81%,3.59%,and 2.33%.(3)For the task of relation recognition,this paper combines the entities and entity location information with the BERT pre-training model to complete the classification of the relationships.In this part,we use the BiLSTM model,BiLSTM-Attention model,and PCNN model as the comparative experiments.The results show that the algorithm in this paper achieves the best results in precision,recall,and F₁ value.(4)This paper constructs the knowledge graph of college student competency and proposes a question-and-answer platform based on it.In this way,we establish a bridge between the competency cultivated by colleges and the competency required for employment positions and provide students with knowledge service based on competency.Finally,this paper illustrates several scenarios in the process of knowledge question-and-answer and presents the results.