Collaborative filtering TopN-recommendation algorithm based on item popularity
Hao Li-ya · Jisuanji gongcheng yu sheji · 2013
To improve the recommendation systems'ability of mining unpopular items,an improved collaborative filtering algorithm is proposed.Based on traditional algorithm,items' popularity is considered as a weighting factor in similarity calculating and recommendation process to boost the reliability of user-similarity calculating and the influence of unpopular items in final recommending.Comparative experiments on typical dataset show that the algorithm is able to mine unpopular items effectively under the premise of maintaining or even improving recommendation accuracy.