Application Of Association Rule Mining In E-commerce Recommendation System

The rapid growth of Internet has caused information overload, particularly in the area of E-commerce, where consumers usually feel troublesome to identify their targets in the deluge of information of potential products. Recommendation systems for E-commerce therefore emerge as the answer to this demand. There has been rapid progress in E-commerce recommendation systems in both fronts of research and practice. However, many E-commerce recommendation systems are still considered immature with problems remained to be solved such as data sparsity, poor scalability and low accuracy.E-commerce recommendation system has gotten great development both in theory and practice. Association rule recommendation technology is currently a more successful personalized recommendation technology in the application of e-commerce recommendation systems. But in practice, there exist such problems as difficulties in discovering association rules and static characteristic of the resulting rules. This paper focuses on these issues and improves it by introducing of the concept hierarchy and establishing evolutionary rule sets. A recommendation model is designed based on evolutionary rule sets.To investigate this model, a personalized recommendation system based on an online toggery is developed on the Myeclipse7 platform. Hit ratio is adopted to evaluate the system performance. Preliminary experimental results show that the accuracy of personalized recommendation is significantly improved.