Cross-modal problem Q matrix automatic construction method based on heterogeneous graph neural network

Abstract

The invention discloses a cross-modal problem Q matrix automatic construction method based on a heterogeneous graph neural network, and the method comprises the steps: constructing a heterogeneous graph containing cross-modal problems and knowledge points at the same time, building the connection between the problems and between the knowledge points according to the similarity, and proposing a heterogeneous graph neural network used for learning the node representation of the heterogeneous graph. And learning node feature representation in the heterogeneous graph and link prediction between problem nodes and knowledge point nodes, and finding out a corresponding relation between problems and knowledge points, so that the purpose of automatically constructing a Q matrix is achieved. The association between the problem and the knowledge point is directly searched in the heterogeneous graph through link prediction in the graph neural network, so that the operation burden of a computer is reduced. Besides, similarity information among the problems is introduced into the prediction process of the knowledge points, and the relation among the problem nodes is increased, so that the accuracy of subsequent knowledge point prediction is improved.

Description

technical field [0001] The invention belongs to the technical field of expert systems, and in particular relates to a method for automatically constructing a cross-modal problem Q matrix. Background technique [0002] Cognitive diagnosis is an important function of the online education system. It can automatically evaluate students' mastery of relevant knowledge points through students' answer scores, help teachers better analyze the deficiencies and defects of students' learning, and enable teachers to understand the importance of teaching. Insufficient and improved, better realize the combination of teaching and evaluation, and promote students to grow more effectively. The Q matrix is ​​composed of the relationship between questions and knowledge points, and it is an important basis for parameter estimation and result prediction of cognitive diagnostic models. [0003] The existing Q-matrix is ​​often manually constructed by experts according to the problem itself, which...

AI Technical Summary

Problems solved by technology

The existing Q matrix automatic construction methods are usually aimed at single-modal problems that only contain text descriptions, and they have the following defects: ①It is not suitable for cross-modal problems

Many methods build the Q matrix based on the student's answer scores. When the student's answer score data is small, many existing methods will lose their accuracy.

Images