Towards Personalized Context-Aware Recommendation by Mining Context Logs through Topic Models

The increasing popularity of smart mobile devices and their more and more powerful sensing ability make it possible to capture rich contextual information and personal context-aware preferences of mobile users by user context logs in devices. By leveraging such information, many context-aware services can be provided for mobile users such as personalized context-aware recommendation. However, to the best knowledge of ours, how to mine user context logs for personalized context-aware recommendation is still under-explored. A critical challenge of this problem is that individual user's historical context logs may be too few to mine their context-aware preferences. To this end, in this paper we propose to mine common context-aware preferences from many users' context logs through topic models and represent each user's personal context-aware preferences as a distribution of the mined common context-aware preferences. The experiments on a real-world data set contains 443 mobile users' historical context data and activity records clearly show the approach is effective and outperform baselines in terms of personalized context-aware recommendation.