Friend Recommendation Algorithm Based on Interest and Cognition Combined with Feedback Mechanism
Yunfei Yin, Xuesong Feng · 2019
Traditional friend recommendation algorithms are mostly based on common friends or similarity of interests. However, when the user's friend relationship or user's interest is sparse, the recommendation result is unsatisfactory. In this case, other factors need to be combined to make recommendations. This paper proposes a method based on the collaborative filtering algorithm that combines interest and cognition to improve the recommendation effect. The mixed similarity is obtained by calculating interest similarity and cognitive similarity and then assigning different weights to them. Also, the traditional friend recommendation algorithm does not take into account the impact of each recommendation result on users. The algorithm proposed in this paper will dynamically adjust the similarity matrix of users based on the results of each recommendation. After each recommendation, we adopt positive feedback adjustment or negative feedback adjustment, which will make the next recommendation more accurate. The experimental results show that compared with the traditional recommendation methods based on common friends or interests, the recommendation algorithm in this paper has a higher precision and recall rate, and the recommendation model can be automatically adjusted according to the user's behavior.