A Location Recommender Based on a Hidden Markov Model: Mobile Social Networks

Abstract

With the development of mobile communication technology and location-based services, people can share information with friends through checking in anywhere, at any time. If we can “speculate” when users will next check in, we can make relevant and useful recommendations. Here, we introduce a new check-in-based hidden Markov model to cope with changing circumstances. A certain check-in-based hidden Markov model for each group is obtained first. The model then analyzes temporal check-in intervals of users before suggesting locations. We also discuss optimal parameter settings for the number of hidden states and the corresponding number of user groups. Experiments show that, given observations of a new entrant, the model is able to predict the most probable time period the user will check in next time. It can also recommend a specific user group for the new entrant. Hence, it enables the recommendation of potential locations of interest for the new entrant.

Keywords:

• check-in prediction
• data sparsity
• hidden Markov model
• location-based services
• mobile social network
• recommendation technique
• social network

Notes

1. ¹http://www.cs.ubc.ca/murphyk/Software/HMM/hmm.html.

Additional information

Notes on contributors

Jialing Li

Jialing Li was a master student at the Department of Computer Science, Southwest University from 2010 through 2013. Her research interests include data mining and big data.

Li Li

Li Li is a full professor in the School of Computer and Information Science at Southwest University in Chongqing, China. She earned her Ph.D. in Computer Engineering from the Swinburne University of Technology, Melbourne, Australia. Her research interests include social networks analysis, data mining, and semantic integration.