TY - GEN
T1 - Mining interesting infrequent and frequent itemsets based on MLMS model
AU - Dong, Xiangjun
AU - Niu, Zhendong
AU - Zhu, Donghua
AU - Zheng, Zhiyun
AU - Jia, Qiuting
PY - 2008
Y1 - 2008
N2 - MLMS (Multiple Level Minimum Supports) model which uses multiple level minimum supports to discover infrequent itemsets and frequent itemsets simultaneously is proposed in our previous work. The reason to discover infrequent itemsets is that there are many valued negative association rules in them. However, some of the itemsets discovered by the MLMS model are not interesting and ought to be pruned. In one of Xindong Wu's papers [1], a pruning strategy (we call it Wu's pruning strategy here) is used to prune uninteresting itemsets. But the pruning strategy is only applied to single minimum support. In this paper, we modify the Wu's pruning strategy to adapt to the MLMS model to prune uninteresting itemsets and we call the MLMS model with the modified Wu's pruning strategy IMLMS (Interesting MLMS) model. Based on the IMLMS model, we design an algorithm to discover simultaneously both interesting frequent itemsets and interesting infrequent itemsets. The experimental results show the validity of the model.
AB - MLMS (Multiple Level Minimum Supports) model which uses multiple level minimum supports to discover infrequent itemsets and frequent itemsets simultaneously is proposed in our previous work. The reason to discover infrequent itemsets is that there are many valued negative association rules in them. However, some of the itemsets discovered by the MLMS model are not interesting and ought to be pruned. In one of Xindong Wu's papers [1], a pruning strategy (we call it Wu's pruning strategy here) is used to prune uninteresting itemsets. But the pruning strategy is only applied to single minimum support. In this paper, we modify the Wu's pruning strategy to adapt to the MLMS model to prune uninteresting itemsets and we call the MLMS model with the modified Wu's pruning strategy IMLMS (Interesting MLMS) model. Based on the IMLMS model, we design an algorithm to discover simultaneously both interesting frequent itemsets and interesting infrequent itemsets. The experimental results show the validity of the model.
UR - https://www.scopus.com/pages/publications/68749083473
U2 - 10.1007/978-3-540-88192-6_42
DO - 10.1007/978-3-540-88192-6_42
M3 - Conference contribution
AN - SCOPUS:68749083473
SN - 3540881913
SN - 9783540881919
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 444
EP - 451
BT - Advanced Data Mining and Applications - 4th International Conference, ADMA 2008, Proceedings
PB - Springer Verlag
T2 - 4th International Conference on Advanced Data Mining and Applications, ADMA 2008
Y2 - 8 October 2008 through 10 October 2008
ER -