Simulating the Effectiveness of Using Association Rules for Recommendation Systems

• Published in: Systems Modeling and Simulation: Theory and Applications pp. 306 - 314
• Main Authors: Chun, Jonghoon, Oh, Jae Young, Kwon, Sedong, Kim, Dongkyu
• Format: Book Chapter
• Language: English
• Published: Berlin, Heidelberg Springer Berlin Heidelberg 2005
• Series: Lecture Notes in Computer Science
• ISBN: 3540244778; 9783540244776
• ISSN: 0302-9743; 1611-3349
• DOI: 10.1007/978-3-540-30585-9_34

Recommendation systems help overcome information overload by providing personalized suggestions based on a history of users’ preference. Association rule-based filtering method is often used for automatic recommendation systems yet it inherently lacks ability to single out a product to recommend for each individual user. In this paper, we propose an association rule ranking algorithm. In the algorithm, we measure how much a user is relevant to every association rule by comparing attributes of a user with the attributes of others who belong to the same association rule. By providing such an algorithm, it is possible to recommend products with associated rankings, which results in better customer satisfaction. We show through simulations, that the accuracy of association rule-based filtering is improved if we appropriately rank association rules for a given user.