Ultra accurate personal recommendation via eliminating redundant correlations

@misc{citeulike:2841716, abstract = {In this paper, based on a weighted projection of bipartite user-object network, we introduce a personal recommendation algorithm which has remarkably higher accuracy than the classical algorithm, namely collaborative filtering. In this algorithm, the correlation resulting from a specific attribute may be repeatedly counted in the cumulative recommendations from different objects. By considering the higher order correlations, we design an effective algorithm that can, to some extent, eliminate the redundant correlations. The algorithmic accuracy, measured by the ranking score, can be further improved by 23\% in the optimal case. Most of the previous studies considered the algorithmic accuracy only, in this paper, we argue that the diversity and popularity, as two significant criteria of algorithmic performance, should also be taken into account. With more or less the same accuracy, an algorithm giving higher diversity and lower popularity is more favorable. Numerical results show that the present algorithm can outperform the standard one simultaneously in all three criteria: higher accuracy, higher diversity, and lower popularity.}, author = {Zhou, Tao and Su, Riqi and Liu, Runran and Jiang, Luoluo and Wang, Bing-Hong and Zhang, Yi-Cheng }, citeulike-article-id = {2841716}, eprint = {0805.4127}, keywords = {collaborative, collaborative-filtering, filtering, recommendation}, month = {May}, posted-at = {2008-06-07 16:15:12}, priority = {2}, title = {Ultra accurate personal recommendation via eliminating redundant correlations}, url = {http://arxiv.org/abs/0805.4127}, year = {2008} }

TY - ELEC ID - citeulike:2841716 L3 - citeulike-article-id:2841716 TI - Ultra accurate personal recommendation via eliminating redundant correlations N2 - In this paper, based on a weighted projection of bipartite user-object network, we introduce a personal recommendation algorithm which has remarkably higher accuracy than the classical algorithm, namely collaborative filtering. In this algorithm, the correlation resulting from a specific attribute may be repeatedly counted in the cumulative recommendations from different objects. By considering the higher order correlations, we design an effective algorithm that can, to some extent, eliminate the redundant correlations. The algorithmic accuracy, measured by the ranking score, can be further improved by 23% in the optimal case. Most of the previous studies considered the algorithmic accuracy only, in this paper, we argue that the diversity and popularity, as two significant criteria of algorithmic performance, should also be taken into account. With more or less the same accuracy, an algorithm giving higher diversity and lower popularity is more favorable. Numerical results show that the present algorithm can outperform the standard one simultaneously in all three criteria: higher accuracy, higher diversity, and lower popularity. KW - collaborative KW - collaborative-filtering KW - filtering KW - recommendation AU - Zhou, Tao AU - Su, Riqi AU - Liu, Runran AU - Jiang, Luoluo AU - Wang, Bing-Hong AU - Zhang, Yi-Cheng PY - 2008/May/27/ UR - http://arxiv.org/abs/0805.4127 ER -