Mining Generalized Association Rules for Service Recommendations for Digital Home Applications

Association rules can be used for service recommendations for digital home applications. Negative associations, which mean the missing of item-sets may imply the appearance of certain item-sets, highlight the implications of the missing item-sets. Many studies have shown that negative associations are as important as the traditional positive ones in practice. The recommendation can be more personalized with the addition of more generalized association rules comprising both positive and negative association rules. In this paper, an algorithm based on the FP-growth framework is proposed to mine the generalized rules. In contrast to previous discovery of negative association rules using the apriori-like approaches, the proposed algorithm efficiently mines the rules and outperforms the apriori-based approach. The algorithm also scales up linearly with the increase of the database size.

[1]  Ramakrishnan Srikant,et al.  Fast algorithms for mining association rules , 1998, VLDB 1998.

[2]  Min Gan,et al.  One extended form for negative association rules and the corresponding mining algorithm , 2005, 2005 International Conference on Machine Learning and Cybernetics.

[3]  Shamkant B. Navathe,et al.  Mining for strong negative associations in a large database of customer transactions , 1998, Proceedings 14th International Conference on Data Engineering.

[4]  Xiaohui Yuan,et al.  Mining negative association rules , 2002, Proceedings ISCC 2002 Seventh International Symposium on Computers and Communications.

[5]  Jian Pei,et al.  Mining frequent patterns without candidate generation , 2000, SIGMOD '00.

[6]  Osmar R. Zaïane,et al.  An associative classifier based on positive and negative rules , 2004, DMKD '04.

[7]  Ming-Syan Chen,et al.  On the mining of substitution rules for statistically dependent items , 2002, 2002 IEEE International Conference on Data Mining, 2002. Proceedings..

[8]  Hongjun Lu,et al.  Exception Rule Mining with a Relative Interestingness Measure , 2000, PAKDD.

[9]  Xindong Wu,et al.  Efficient mining of both positive and negative association rules , 2004, TOIS.