Abstract | ||
---|---|---|
Due to the rapid development of E-commerce, personalized recommendations have been indispensable. The conventional user-based collaborative filtering CF cannot well satisfy users' requirements, besides the recommendation results are not accurate enough. To improve the conventional user-based CF, this paper proposes an adaptive CF method based on multiple features. We take four considerations into account: 1 redefining itemitem/ user-user similarity by utilizing item/user vector; 2 making predictions based on the relation between the predicted item and the rated similar items; 3 modifying the rating according to the interest in the type of item; 4 improving the diversity of recommendation. The proposed method is easy to implement, and experimental results based on two well-known datasets have demonstrated the superiority in accuracy and diversity. |
Year | DOI | Venue |
---|---|---|
2013 | 10.1007/978-3-642-53917-6_41 | ADMA |
Keywords | Field | DocType |
cf,item recommendation,item-item similarity,user-user similarity | Data mining,Collaborative filtering,Information retrieval,Computer science,Artificial intelligence,Machine learning | Conference |
Volume | Issue | Citations |
8347 LNAI | PART 2 | 0 |
PageRank | References | Authors |
0.34 | 21 | 3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Zhang Yan-Qiu | 1 | 0 | 0.34 |
Zheng Hai-Tao | 2 | 142 | 24.39 |
Zhang Lan-Shan | 3 | 2 | 0.69 |