A naive recommendation model for large databases

Abstract

It is difficult for users to find items as the number of choices increase and they become overwhelmed with high volume of data. In order to avoid them from bewilderment, a recommender could be applied to find more related items in shorter time. In this paper, we proposed a naive recommender model which uses Association Rules Mining technique to generate two item sets enabling to find all existing rules for a certain item and has the capability to search on demand which decrease the response time dramatically This model mines transactions’ database to discover the existing rules among items and stores them in a sparse matrix. It also searches the matrix by means of a naive algorithm to generate a search list.We have applied and evaluated our model in Universiti Teknologi Malaysia and the results reflect a high level of accuracy.