S. Chaudhury et al. (Eds.): PReMI 2009, LNCS 5909, pp. 255–260, 2009. © Springer-Verlag Berlin Heidelberg 2009 Mining Local Association Rules from Temporal Data Set Fokrul Alom Mazarbhuiya 1 , Muhammad Abulaish 2,∗ , Anjana Kakoti Mahanta 3 , and Tanvir Ahmad 4 1 College of Computer Science, King Khalid University, Abha, KSA fokrul_2005@yahoo.com 2 Department of Computer Science, Jamia Millia Islamia, Delhi, India abulaish@ieee.org 3 Department of Computer Science, Gauhati University, Assam, India anjanagu@yahoo.co.in 4 Department of Computer Engineering, Jamia Millia Islamia, Delhi, India tanvir.ce@jmi.ac.in Abstract. In this paper, we present a novel approach for finding association rules from locally frequent itemsets using rough set and boolean reasoning. The rules mined so are termed as local association rules. The efficacy of the pro- posed approach is established through experiment over retail dataset that con- tains retail market basket data from an anonymous Belgian retail store. Keywords: Data mining, Temporal data mining, Local association rule mining, Rough set, Boolean reasoning. 1 Introduction Mining association rules in transaction data is a well studied problem in the field of data mining. In this problem, given a set of items and a large collection of transac- tions, the task is to find relationships among items satisfying a user given support and confidence threshold values. However, the transaction data are temporal in the sense that when a transaction happens the time of transaction is also recorded. Considering the time aspect, different methods [1] have been proposed to extract temporal associa- tion rules, i.e., rules that hold throughout the life-time of the itemset rather than throughout the life-time of the dataset. The lifetime of an itemset is the time period between the first transaction containing the itemset and the last transaction containing the same itemset in the dataset and it may not be same as the lifetime of the dataset. Mahanta et al. have addressed the problem of temporal association rule extraction in [2]. They proposed an algorithm for finding frequent itemsets with respect to a small time-period not necessarily equal to the lifetime of the dataset or that of the itemset. They named such itemsets as locally frequent itemsets and corresponding rules as local association rules. In order to calculate the confidence value of a local associa- tion rule, say A ⇒ X – A, in the interval [t, t′] where X is a frequent itemset in [t, t′] and X A ⊂ , it is required to know the supports of both X and A in the same interval ∗ To whom correspondence should be addressed.