Call for
for Authors
Workshop Program Program Committee Past SNAKDD workshops

Accepted Papers    

    Full Paper:

  • Finding Contexts of Social Influence in Online Social Networks (Student Travel Award)
    Jennifer H. Nguyen, Bo Hu, Stephan GŁnnemann and Martin Ester
  • ProfileRank: Finding Relevant Content and Influential Users based on Information Diffusion (Student Travel Award)
    Arlei Silva, Sara Guimar„es, Wagner Meira Jr. and Mohammed Zaki
  • Network Flows and the Link Prediction Problem
    Kanika Narang, Kristina Lerman and Ponnurangam Kumaraguru
  • Epidemiological Modeling of News and Rumors on Twitter (Student Travel Award)
    Fang Jin, Edward Doughertyu, Parang Saraf, Yang Cao and Naren Ramakrishnan
  • Modeling Direct and Indirect Influence across Heterogeneous Social Networks
    Minkyoung Kim, David Newth and Peter Christen
  • Structure and Attributes Community Detection: Comparative Analysis of Composite, Ensemble and Selection Methods
    Haithum Elhadi and Gady Agam

  • Short Paper:

  • Community Finding within the Community Set Space
    Jerry Scripps and Christian Trefftz
  • Analysis and Identification of Spamming Behaviors in Sina Weibo Microblog
    Chengfeng Lin, Yi Zhou, Kai Chen, Jianhua He, Li Song and Xiaokang Yang
  • CUT: Community Update and Tracking in Dynamic Social Networks
    Hao-Shang Ma and Jen-Wei Huang
  • Leveraging Candidate Popularity On Twitter To Predict Election Outcome
    Manish Gaurav, Anoop Kumar, Amit Srivastava and Scott Miller
  • Mixing Bandits: A Recipe for Improved Cold-Start Recommendations in a Social Network
    Stťphane Caron and Smriti Bhagat
  • The User's Communication Patterns on A Mobile Social Network Site
    Youngsoo Kim
  • Twitter Volume Spikes: Analysis and Application in Stock Trading
    Yuexin Mao, and Wei Wei and Bing Wang
  • Customized Reviews for Small User-Databases using Iterative SVD and Content Based Filtering
    Jon Gregg and Nitin Jain
Student Travel Awards    
  • Epidemiological Modeling of News and Rumors on Twitter
    - *Fang Jin (Virginia Tech, Blacksburg)
    - Edward Dougherty (Virginia Tech, Blacksburg)
    - Parang Saraf (Virginia Tech, Blacksburg)
    - Yang Cao (Virginia Tech, Blacksburg)
    - Naren Ramakrishnan (Virginia Tech, Blacksburg)
  • Abstract: Characterizing information diffusion on social platforms like Twitter enables us to understand the properties of underlying media and model communication patterns. As Twitter gains in popularity, it has also become a venue to broadcast rumors and misinformation. We use epidemiological models to characterize information cascades in twitter resulting from both news and rumors. Specifically, we use the SEIZ enhanced epidemic model that explicitly recognizes skeptics to characterize eight events across the world and spanning a range of event types. We demonstrate that our approach is accurate at capturing diffusion in these events. Our approach can be fruitfully combined with other strategies that use content modeling and graph theoretic features to detect (and possibly disrupt) rumors.

  • Finding Contexts of Social Influence in Online Social Networks
    - Jennifer H. Nguyen(University College London, UK)
    - Bo Hu(Simon Fraser University, Canada)
    - *Stephan GŁnnemann(Carnegie Mellon University, USA)
    - Martin Ester(Simon Fraser University, Canada)

    Abstract: The ever rising popularity of online social networks has not only attracted much attention from everyday users but also from academic researchers. In particular, research has been done to investigate the effect of social in uence on users' actions on items in the network. However, all social in uence research in the data-mining field has been done in a contextindependent setting, i.e., irrespective of an item's characteristics. It would be interesting to find the specific contexts in which users in uence each other in a similar manner. In this way, applications such as recommendation engines can focus on a specific context for making recommendations. In this paper, we pose the problem of finding contexts of social in uence where the social in uence is similar across all items in the context. We present a full-space clustering algorithm and a subspace clustering algorithm to find these contexts and test the algorithms on the Digg data set. We demonstrate that our algorithms are capable of finding meaningful contexts of in uence in addition to rediscovering the predefined categories specific to the Digg news site.

  • ProfileRank: Finding Relevant Content and Influential Users based on Information Diffusion
    - *Arlei Silva (Universidade Federal de Minas Gerais)
    - Sara Guimar„es (Universidade Federal de Minas Gerais)
    - Wagner Meira Jr. (Universidade Federal de Minas Gerais)
    - Mohammed Zaki (Rensselaer Polytechnic Institute)

    Abstract: Understanding information diffusion processes that take place on the Web, specially in social media, is a fundamental step towards the design of effective information diffusion mechanisms, recommendation systems, and viral marketing/advertising campaigns. Two key concepts in information diffusion are influence and relevance. Influence is the ability to popularize content in an online community. To this end, influentials introduce and propagate relevant content, in the sense that such content satisfies the information needs of a significant portion of this community. In this paper, we study the problem of identifying in?uential users and relevant content in information diffusion data. We propose ProfileRank, a new information diffusion model based on random walks over a user-content graph. ProfileRank is a PageRank inspired model that exploits the principle that relevant content is created and propagated by in?uential users and influential users create relevant content. A convenient property of ProfileRank is that it can be adapted to provide personalized recommendations. Experimental results demonstrate that ProfileRank makes accurate recommendations, outperforming baseline techniques. We also illustrate relevant content and influential users discovered using ProfileRank. Our analysis shows that ProfileRank scores are more correlated with content diffusion than with the network structure. We also show that our new modeling is more efficient than PageRank to perform these calculations.