PROJECT TITLE :

Author Topic Model-Based Collaborative Filtering for Personalized POI Recommendations

ABSTRACT:

From social media has emerged continuous desires for automatic travel recommendations. Collaborative filtering (CF) is the foremost well-known approach. But, existing approaches usually suffer from numerous weaknesses. For example , sparsity can considerably degrade the performance of ancient CF. If a user only visits terribly few locations, correct similar user identification becomes very challenging due to lack of sufficient data for effective inference. Moreover, existing recommendation approaches typically ignore wealthy user information like textual descriptions of photos that will replicate users’ travel preferences. The topic model (TM) methodology is an effective way to solve the “sparsity problem,” however remains far from satisfactory. During this paper, an author topic model-based mostly collaborative filtering (ATCF) methodology is proposed to facilitate comprehensive points of interest (POIs) recommendations for social users. In our approach, user preference topics, like cultural, cityscape, or landmark, are extracted from the geo-tag constrained textual description of photos via the author topic model rather than solely from the geo-tags (GPS locations). Benefits and superior performance of our approach are demonstrated by intensive experiments on a large collection of knowledge.