TY - GEN
T1 - Mining top-K association rules
AU - Fournier-Viger, Philippe
AU - Wu, Cheng Wei
AU - Tseng, Vincent S.
PY - 2012/6/6
Y1 - 2012/6/6
N2 - Mining association rules is a fundamental data mining task. However, depending on the choice of the parameters (the minimum confidence and minimum support), current algorithms can become very slow and generate an extremely large amount of results or generate too few results, omitting valuable information.This is a serious problem because in practice users have limited resources for analyzing the results and thus are often only interested in discovering a certain amount of results, and fine tuning the parameters is time-consuming.To address this problem, we propose an algorithm to mine the top-k association rules, where k is the number of association rules to be found and is set by the user. The algorithm utilizes a new approach for generating association rules named rule expansions and includes several optimizations. Experimental results show that the algorithm has excellent performance and scalability, and that it is an advantageous alternative to classical association rule mining algorithms when the user want to control the number of rules generated.
AB - Mining association rules is a fundamental data mining task. However, depending on the choice of the parameters (the minimum confidence and minimum support), current algorithms can become very slow and generate an extremely large amount of results or generate too few results, omitting valuable information.This is a serious problem because in practice users have limited resources for analyzing the results and thus are often only interested in discovering a certain amount of results, and fine tuning the parameters is time-consuming.To address this problem, we propose an algorithm to mine the top-k association rules, where k is the number of association rules to be found and is set by the user. The algorithm utilizes a new approach for generating association rules named rule expansions and includes several optimizations. Experimental results show that the algorithm has excellent performance and scalability, and that it is an advantageous alternative to classical association rule mining algorithms when the user want to control the number of rules generated.
KW - association rule mining
KW - rule expansion
KW - support
KW - top-k rules
UR - http://www.scopus.com/inward/record.url?scp=84861752193&partnerID=8YFLogxK
U2 - 10.1007/978-3-642-30353-1_6
DO - 10.1007/978-3-642-30353-1_6
M3 - Conference contribution
AN - SCOPUS:84861752193
SN - 9783642303524
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 61
EP - 73
BT - Advances in Artificial Intelligence - 25th Canadian Conference on Artificial Intelligence, Canadian AI 2012, Proceedings
T2 - 25th Canadian Conference on Artificial Intelligence, AI 2012
Y2 - 28 May 2012 through 30 May 2012
ER -