Graph-based metric embedding for next POI recommendation

Xie, Min, Yin, Hongzhi, Xu, Fanjiang, Wang, Hao and Zhou, Xiaofang (2017). Graph-based metric embedding for next POI recommendation. In: Wojciech Cellary, Mohamed F. Mokbel, Jianmin Wang, Hua Wang, Rui Zhou and Yanchun Zhang, Web Information Systems Engineering – WISE 2016. 17th International Conference on Web Information Systems Engineering (WISE), Shanghai, China, (207-222). 8 - 10 November 2016. doi:10.1007/978-3-319-48743-4_17

Author Xie, Min
Yin, Hongzhi
Xu, Fanjiang
Wang, Hao
Zhou, Xiaofang
Title of paper Graph-based metric embedding for next POI recommendation
Conference name 17th International Conference on Web Information Systems Engineering (WISE)
Conference location Shanghai, China
Conference dates 8 - 10 November 2016
Proceedings title Web Information Systems Engineering – WISE 2016   Check publisher's open access policy
Journal name Lecture Notes in Computer Science   Check publisher's open access policy
Series Lecture Notes in Computer Science
Place of Publication Heidelberg, Germany
Publisher Springer
Publication Year 2017
Sub-type Fully published paper
DOI 10.1007/978-3-319-48743-4_17
Open Access Status Not yet assessed
ISBN 9783319487427
ISSN 1611-3349
Editor Wojciech Cellary
Mohamed F. Mokbel
Jianmin Wang
Hua Wang
Rui Zhou
Yanchun Zhang
Volume 10042
Start page 207
End page 222
Total pages 16
Language eng
Abstract/Summary With the rapid prevalence of smart mobile devices and the dramatic proliferation of location-based social networks (LBSNs), point of interest (POI) recommendation has become an important means to help people discover attractive and interesting places. In this paper, we investigate the problem of next POI recommendation by considering the sequential influences of POIs, as a natural extension of the general POI recommendation, but it is more challenging than the general POI recommendation, due to that (1) users’ preferences are dynamic, and the next POI recommendation requires tracking the change of user preferences in a real-time manner; and (2) the prediction space is extremely large, with millions of distinct POIs as the next prediction target, which impedes the application of classical Markov chain models. In light of the above challenges, we propose a graph-based metric embedding model which converts POIs in a low dimensional metric and tracks the dynamics of user preferences in an efficient way. Besides, the knowledge of sequential patterns of users’ check-in behaviors can be exploited and encoded in the POI embedding, which avoid the time-consuming computation of the POI-POI transition matrix or even cube as the Markov chain-based recommender models have done. In other words, our proposed method effectively unifies dynamic user preferences and sequential influence via the POI embedding. Experiments on two real large-scale datasets demonstrate a significant improvement of our proposed models in terms of recommendation accuracy, compared with the state-of-the-art methods.
Keyword Dynamic user preferences
Metric embedding
Next POI recommendation
Sequential influence
Q-Index Code E1
Q-Index Status Provisional Code
Institutional Status UQ

Version Filter Type
Citation counts: TR Web of Science Citation Count  Cited 0 times in Thomson Reuters Web of Science Article
Scopus Citation Count Cited 1 times in Scopus Article | Citations
Google Scholar Search Google Scholar
Created: Sun, 19 Mar 2017, 01:00:25 EST by System User on behalf of Learning and Research Services (UQ Library)