Learning recommendation based on hybrid collaborative filtering algorithm

Abstract

To solve the sparse matrix problem of learning resource recommendation algorithm based on collaborative filtering, a hybrid recommendation algorithm combining user learning behaviour and rating is proposed. First of all, learning behaviour characteristics are defined and different characteristics are given weight values; then, according to the user’s learning behaviour log, the method of weight accumulation to calculate the user’s learning behaviour value of resources is used; finally, the learning behaviour matrix and rating matrix is used to build a hybrid collaborative filtering recommendation algorithm. Experimental results show that the hybrid recommendation algorithm has lower root mean square error than the traditional collaborative filtering algorithm.