Interactive Social Recommendation

摘要: Social recommendation has been an active research topic over the last decade, based on assumption that social information from friendship networks is beneficial for improving accuracy, especially when dealing with cold-start users who lack sufficient past behavior accurate recommendation. However, it nontrivial to use such information, since some of a person's friends may share similar preferences in certain aspects, but others be totally irrelevant recommendations. Thus one challenge explore and exploit extend which user trusts his/her utilizing improve On other hand, most existing models are non-interactive their algorithmic strategies batch learning methodology, learns train model offline manner collection training data accumulated users? historical interactions recommender systems. In real world, new leave systems reason being recommended boring items before enough collected good model, results inefficient customer retention. To tackle these challenges, we propose novel method interactive recommendation, not only simultaneously explores exploits effectiveness personalization way, also adaptively different weights friends. addition, give analyses complexity regret proposed model. Extensive experiments three real-world datasets illustrate improvement our against state-of-the-art algorithms.