A sequence recommendation method considering course leading relationship and course classification

Abstract

The application discloses a sequence recommendation method considering course leading relationship and course classification, comprising the following steps: 1, constructing a knowledge graph according to the interaction data of users and courses to capture the leading relationship between courses; 2, modeling the interest of users at the course level by using a gated graph neural network and an attention mechanism according to the historical course sequence of user learning; 3, modeling the long-term interest transfer of users at the classification level by using a GRU according to the historical course sequence of user learning and the course classification information; 4, combining the two interests of users and the leading relationship between courses to predict the preference of users for courses, and selecting a suitable loss function to optimize the model; 5, predicting the preference of users for courses by using the recommendation model, and recommending the next course for users. The application combines the knowledge graph and the graph neural network, can capture the leading relationship between courses, and can capture the interest change of users from the course level and the classification level, so that more accurate recommendation effect is realized.

Images