Title: Using a cosine-type measure to derive strong association mining rules

Authors: Sikha Bagui, Jiri Just, Subhash C. Bagui, Rohan Hemasinha

Addresses: Department of Computer Science, University of West Florida, Pensacola, FL 32514, USA. ' Department of Computer Science, University of West Florida, Pensacola, FL 32514, USA. ' Department of Mathematics and Statistics, University of West Florida, Pensacola, FL 32514, USA. ' Department of Mathematics and Statistics, University of West Florida, Pensacola, FL 32514, USA

Abstract: Association mining rule algorithms have two major drawbacks – the need to repeatedly scan the dataset and the generation of too many association rules. In this paper we present an algorithm that concentrates on addressing these drawbacks. We present a correlation based association mining rule algorithm, implemented using an arraylist structure in JAVA, that does not require more than one scan of the full dataset and generates far lot less strong association mining rules. The correlation criteria used is a cosine-type measure.

Keywords: association rule mining; Apriori algorithm; strong association rules; correlation measures; cosine; association mining.

DOI: 10.1504/IJKEDM.2010.032581

International Journal of Knowledge Engineering and Data Mining, 2010 Vol.1 No.1, pp.69 - 83

Available online: 08 Apr 2010 *

Full-text access for editors Access for subscribers Purchase this article Comment on this article