A Simple and Heuristic Model of Tag Recommendation

Abstract

Compared to the high computational complexity of many tag recommenders, a simple and heuristic approach of tag recommendation is proposed, based on tag user and item tag co-occurrences in parallel. Firstly, we use aspect model PLSA to set up a probabilistic model. We find that the probability of the recommended tags to an item for a specific user is determined by two factors: the preferences in choosing tags for the user and the tags reflecting the feature of the item. Then we immerge the two factors into a unified representation. The experiments show that our approach not only has better reliability and precision, but also is very simple and more practical than other algorithms.