Collaborative filtering algorithm based on user trust and interest drift detecting

To solve this problem, the collaborative filtering algorithm based on user trust and user interest CF-BI was proposed. Firstly, according to the user's history score matrix, the trust model of the comprehensive user preference similarity and user credibility was proposed, which took full account of user preference similarity, user influence and scoring professional and other influencing factors; and then the similarity of users' interests was calculated by Pearson correlation coefficient of Ebbinghaus forgetting function, and the similarity degree and interest similarity degree between users were adjusted by the weighting coefficient, which made the selection of the nearest neighbor more accurate, and recommends the Top-N algorithm to the target users. Experimental results on the MovieLens dataset show that the average absolute error of the algorithm is 16.98% than that of the traditional collaborative filtering algorithm, which improves the quality of recommendation effectively.