A new Tag Recombinant Approach Based on Apriori Algorithm

In the recommender system, users often label multiple same tags on the interesting items which other users have labeled. It can cause that many same tags are often repeatedly labeled, resulting in the tag redundancy situation on the user-item-tag data set. In this paper, a tag recombinant approach based on Apriori algorithm is proposed for reducing the tag redundancy situation. The approach first utilized the Apriori algorithm to preprocess the original tag data for searching the frequent itemsets of tags, and then recombined the new tags to form a new data set. The experimental results demonstrated that based on the proposed approach, the newly formed data set significantly reduced the number of the tuples, and at the same time the relationship among the user, item and tag can be reflected more clearly.