CIKM 2017 Tutorial: Towards Space and Time Coupled Social Media Analysis


With the prevalence of versatile mobile devices, people’s offline activities are being increasingly captured and shared via online social media. Every day, billions of people probe different places in the physical world and broadcast their activities on various platforms (e.g., Facebook, Twitter, Instagram, Yelp) in the form of geo-tagged social media posts. The confluence of multimodal information (location, time, and text) in such data offers new opportunities for extracting valuable knowledge about people’s activities, but meanwhile also introduces unique challenges to conventional data mining techniques. In the past few years, a large body of space and time coupled social media analysis methods have emerged to model people’s activities in rich contexts and have been shown to be powerful in improving downstream tasks. In this tutorial, we present an organized picture of existing techniques for space and time coupled social media analysis, covering topics including spatiotemporal activity mining, event detection and forecasting, mobility modeling, and location recommendation and prediction. We also discuss about the limitations of existing research as well as important future directions. We believe this tutorial will be of interest to both researchers and practitioners in this field.


  • Chao Zhang is a Ph.D. candidate at the Department of Computer Science, University of Illinois at Urbana-Champaign. His research focuses on knowledge discovery from social media data and multimodal data mining. He has won the 2015 ECML/PKDD Best Student Paper Runner-up Award, the Microsoft Star of Tomorrow Excellence Award, and the Chiang Chen Overseas Graduate Fellowship.

  • Quan Yuan is a research scientist at Facebook Inc. He received his Ph.D. degree from the School of Computer Engineering, Nanyang Technological University, Singapore in 2015. He was a postdoctoral research associate of the Department of Computer Science at University of Illinois at Urbana-Champaign from 2015 to 2017. His research interests include spatio-temporal data mining, recommender systems, and text mining.

  • Shi Zhi is a Ph.D. candidate at the Department of Computer Science, University of Illinois at Urbana-Champaign. She received her B.S. degree from Electronic Engineering Department of Tsinghua University in 2012. Her research interests are focused on truth discovery, graph mining, text mining and spatial-temporal mining.

  • Sha Li is a master student at the Department of Computer Science, University of Illinois at Urbana-Champaign. She received her bachelor degree from the Department of Computer Science, Shanghai Jiao Tong University. Her current research interests are in user behavior analysis and text mining.

  • Jiawei Han is an Abel Bliss Professor at the Department of Computer Science, UIUC. His research areas encompass data mining, data ware-housing, information network analysis, and database systems, with over 600 conference and journal publications. He is Fellow of ACM and Fellow of IEEE, and received ACM SIGKDD Innovation Award (2004), IEEE Computer Society Technical Achievement Award (2005), and IEEE Computer Society W. Wallace McDowell Award (2009). His co-authored textbook “Data Mining: Concepts and Techniques”, 3rd ed., (Morgan Kaufmann, 2011) has been adopted popularly world-wide.


  • Introduction (15 min)
    • Motivations
      • Conventional social media analysis
      • The emergence of geo-tagged social media
      • Opportunities of incorporating rich context
    • Overview of space and time coupled social media analysis
      • Challenges
      • Categorization of the tasks
      • General principles Downstream applications
    • QA session
  • Spatiotemporal Acvitiy Modeling (30 min)
    • Similarity-based methods
    • Probabilistic graphical modeling methods
    • Representation learning methods
    • QA session
  • Event Detection and Forecasting (40 min)
    • Global event detection
      • Feature-based methods
      • Document-based methods
    • Local event detection
      • Batch detection
      • On-line detection
    • Event forecasting by combining social media with other sources
    • QA session
  • Mobility Modeling (40 min)
    • Mobility pattern mining
      • Sequential mobility patterns
      • Periodic mobility patterns
    • Model-based approaches
      • Hidden Markov model methods
      • Recurrent neural network methods
    • QA session
  • Location Recommendation and Prediction (40 min)
    • Location recommendation
      • User-based collaborative filtering
      • Matrix factorization models
      • Poisson factor models
    • Location prediction
      • Frequent pattern-based models
      • Hidden Markov models
      • Supervised ranking models
    • QA session
  • Summary and Future Directions (15 min)
    • Summary of space and time coupled social media analysis
      • Principles and techniques
      • Advantages and limitations
      • How to choose a method based on your application?
    • Future directions
      • Handling data sparsity
      • Scalable and on-line mining
      • Representation learning for space and time coupled social media
      • QA session


Available here