Multi Agent Collaborative Filtering Algorithm Based on Double Layer Behavior Driven

Abstract

In recent years, with the development of the Internet and network learning community rapid development, information technology is changing people's ways of e-learning with amazing speed. At the same time, learning trek, knowledge overload problems gradually emerged. Personalized learning recommendation has gradually become an important means for people to quickly learn and master the knowledge, but also reflects the people-oriented personalized learning mode. However, recommendation algorithm based on collaborative filtering still faces problems such as sparsity, scalability, cold start and precision. In this paper, a multi Agent collaborative filtering model based on double layer behavior is proposed for the problem of data sparsity and accuracy. Using social network relationship and dynamic agent perception and the ability to interact with learning, combined with individual behavior to measure the degree of interest, group behavior measure of trust and influence of learning resources for collaborative forecasting scale, reduce the score prediction error and the problem of learning trek.