Publications

Point-of-Interest Recommendation Using Heterogeneous Link Prediction

Alireza Pourali and Fattane Zarrinkalam and Ebrahim Bagheri
Reference:
Alireza Pourali; Fattane Zarrinkalam and Ebrahim Bagheri Point-of-Interest Recommendation Using Heterogeneous Link Prediction. In The International Conference on Extending Database Technology (EDBT), March 26-29, Vienna, Austria, 2018.
Links to Publication:
Abstract:
Venue recommendation in location-based social networks is among the more important tasks that enhances user participation on the social network. Despite its importance, earlier research have shown that the accurate recommendation of appropriate venues for users is a difficult task specially given the highly sparse nature of user check-in information. In this paper, we show how a comprehensive set of user and venue related information can be methodically incorporated into a heterogeneous graph representation based on which the problem of venue recommendation can be viewed as an instance of the heterogeneous link prediction over the graph. We systematically compare our proposed approach with several strong baselines and show that our work that is computationally less-intensive compared to the baselines shows improved performance in terms of precision and f-measure.
Bibtex Entry:
@inproceedings{edbt2018, author = {Alireza Pourali and Fattane Zarrinkalam and Ebrahim Bagheri}, title = {Point-of-Interest Recommendation Using Heterogeneous Link Prediction}, booktitle = {The International Conference on Extending Database Technology (EDBT), March 26-29, Vienna, Austria}, abstract = {Venue recommendation in location-based social networks is among the more important tasks that enhances user participation on the social network. Despite its importance, earlier research have shown that the accurate recommendation of appropriate venues for users is a difficult task specially given the highly sparse nature of user check-in information. In this paper, we show how a comprehensive set of user and venue related information can be methodically incorporated into a heterogeneous graph representation based on which the problem of venue recommendation can be viewed as an instance of the heterogeneous link prediction over the graph. We systematically compare our proposed approach with several strong baselines and show that our work that is computationally less-intensive compared to the baselines shows improved performance in terms of precision and f-measure.}, year = {2018} }




Powered by WordPress