Search

Recommendation methods are becoming more and more important for the online shopping websites. The development of multi-criteria recommender systems such as initiator recommendation according to customers 'preferences have been a noteworthy decision problem. Also, it is interesting to study how to design initiator recommender systems for group-buying. Whereas grey relational analysis (GRA) is a useful technique that can assess the relationships among patterns, this study presents an aggregation-function-based method by using the proposed grey criterion similarity. The proposed similarity measure reveals that the greater the strength of the relationship of one user with another one for a criterion, the greater criterion similarity of the former to the latter Besides, the aggregation-function-based method can be used to predict unknown overall ratings by using the proposed criterion similarity to estimate multi-criteria ratings. Experimental results on a group-buying website have demonstrated that the generalization ability of the aggregation-function-based method using the proposed grey criterion similarity performs well in comparison to that using other criterion similarities. [ABSTRACT FROM AUTHOR]

Previously, we proposed a novel similarity measure for neighborhood methods using the grey relational analysis (GRA). Nevertheless, the traditional GRA have drawn our attention to its assumption on noninteraction among criteria involved. Since criteria are not always independent, the adequacy of additivity should be taken into account. This motivates us to use the nonadditive GRA to design a novel similarity measure for neighborhood methods in collaborative filtering on the basis of the previous study. The neighborhood methods are then used to recommend items that users with similar preferences liked in the past. The applicability of the proposed fuzzy similarity-based method to movie recommendation is examined. Experimental results on movie recommendation demonstrate that the generalization ability of the multi-criteria neighborhood method using the proposed fuzzy similarity measure performs well in comparison to that using the additive similarity measure on the basis of GRA. [ABSTRACT FROM AUTHOR]