融合协同过滤的自组织神经网络多样化产品推荐OA北大核心CSTPCD

Aiming at the redundancy problem of the recommendation results for personalized recommendation algorithms,a diversi-fied product recommendation method,integrating collaborative filtering with self-organizing neural network,is proposed.Firstly,the user-product rating table and user-product category rating table are constructed through the user's rating of the product.Furtherly,we adopt the collaborative filtering algorithm to get the product recommendation list based on similar users with similar ratings.Second-ly,the user vectors are input into the self-organizing neural network to cluster similar users,and the similar users are used to help se-lect the product categories that the target user may be interested in.A diversified recommendation list is formed.Finally,the two rec-ommendation lists are fused to construct the diversified and accurate product recommendation results.The experiments on the Ama-zon datasets verified that the proposed method reaches better results on category coverage(CC)and item-level diversity(ILD)index-es and can effectively carry out diversified recommendations.