Mining positive and negative weighted association rules in medical records without user-specified weights based on HITS model

One of the challenging problems in the weighted association rules mining is to assign weights to items. For practice, self-assigned weights technique is more useful. In this paper, we proposed a self-assigned weights method to discover positive and negative association rules, instead of assigning the weights by users. To avoid mining misleading and uninteresting rules, a new type parameter, called sawinterest, is proposed to eliminate the redundant rules. The rational results are presented.