Research on Curriculum Resource-oriented Fusion Knowledge Graph and Multi-task Feature Recommendation Algorithm

Authors

  • Wu Hao , Xu Xingjian , Meng Fanjun

Abstract

Since some data are unavailable and missing when collecting course data, resulting in the
generation of missing values in the data, making data sparsity and cold start problems when
recommending course resources, researchers usually use additional information to mitigate the
problem of reduced accuracy due to missing values. In this paper, a recommendation algorithm
with more accurate output results is proposed based on a multi-task feature learning approach,
using the course knowledge graph as auxiliary information. The algorithm is based on an endto-end framework for deep learning, with the knowledge graph embedded in the tasks to assist
in recommendation, and the tasks are correlated with each other through cross-compression
units, and automatically share potential features, learn items in the recommendation algorithm
and higher-order connections between entities in the knowledge graph to build a
recommendation model.

Published

2020-10-15

Issue

Section

Articles