CSS

From PgroupW

Jump to: navigation, search

Computational Social Science (CSS) Team

Data availability in online social networks as well as the business world has lately not been an issue. On the contrary, vast amounts of data are being generated by social networking users in the form of informal interactions. What has been an issue however, is the transformation of data into useful information, that in time and with appropriate processing becomes knowledge. In this research project, we examine knowledge generation under informal social communications, based on semantically enriched user generated data. Modeling users with their generated content, we are able to dynamically capture users’ interests. Knowledge networks of users emerge, exhibiting collective intelligence. To capture such collective knowledge, we propose a novel knowledge base paradigm, which seamlessly integrates information from multiple platforms and facilitates knowledge extraction, mining, discovery and inferencing. We use such integrated, semantically rich information to calculate semantic similarity between people under the context of informal social interactions, driving numerous applications.

Research Topics

  • Social Network Analytics
  • Link Prediction
  • Sentiment Analysis
  • Opinion Mining
  • Enterprise Knowledge Management
  • Graph Mining


Research Projects

Predicting Communication Intention

Predicting Communication intention
Predicting Communication intention

In social networks, where users send messages to each other, the issue of what triggers communication between unrelated users arises: does communication between previously unrelated users depend on FOAF-like relationships, common interests, or other factors? In this work, we study the problem of predicting directed communication intention between two users. We propose a new approach that jointly considers local structural properties of users in a social network, in conjunction with their generated content. We evaluate our method using an extracted network of directed @-messages sent between users of a corporate microblogging service, which resembles Twitter.

Microblogging Behavior in the Enterprise

Empirical Analysis Microblogging
Empirical Analysis Microblogging: Cumulative umber of replies

Popular social networking sites have revolutionized the way people interact on the Web. Researchers have studied social networks from numerous perspectives, mostly focusing on publicly available social networks and microblogging sites. Enterprises however have recently being adopting and utilizing microblogging services as part of their day to day operations. The goal of this research project is to study the topological properties of a corporate microblogging service, its dynamics and characteristics, and the interplay between its social and topical components. Through an extensive analysis of enterprise microblogging data, we provide insights on the structural properties of the extracted network of directed messages sent between users of a corporate microblogging service, as well as the lexical and topical alignment of users. We study homophily, finding a substantial level of alignment with respect to the network structure and users activity, as well as latent topical similarity and link probability. Our analysis suggests that users with strong local topical alignment tend to participate in focused interactions, whereas users with disperse interests contribute to multiple discussions, broadening the diversity of articipants. We compare our results to traditional, general purpose, online social networks and discuss the implications of our findings. To the best of our knowledge, this work is the first quantitative study of an enterprise microblogging service, its usage characteristics, and its derived social network based on replies between users.


Enterprise knowledge preservation and management

Decision making process in today’s organizations is constantly evolving with expanding geographical boundaries and ever changing technology landscape. Major part of decisions and deliberations are now typically taking place in collaboration platforms like enterprise social networks, emails, discussion servers, chat, and conferencing services. As these platforms contain problem solving insights, recommendations, best practices, expert opinions, and answers; they must be considered part of organizational knowledge management effort. However, traditional knowledge management techniques do not sufficiently capture the hidden nuggets of knowledge buried in communication logs. In this research project, we introduce need of paradigm shift in knowledge management strategy, and discuss the role of semantic social network analysis as a potential solution. We introduce the concept of social knowledge network and describe knowledge algebra by defining rigorous social metrics. Finally, to demonstrate applicability of this approach, we provide two case studies that lead to identification of experts and best practices from organizational communication.

Semantic Social Network analysis for the enterprise

Dimensions in Social Network Analytics
Dimensions in Social Network Analytics

Business processes are generally fixed and enforced strictly. This is reflected in static nature of underlying software systems and datasets, thereby resulting in resistance to change. However, internal and external situation, organizational changes and various other factors triggers dynamism. Such dynamism is generally observed in communication channels in the form of issues, complains, Q&A, opinion, review etc. These channels can be one or mix of email, chat, discussion forum, and official communication, internal social network. Careful and timely analysis and processing of such channels may lead to early detection of emerging trends, critical issues, opportunities, topics of interests, contributors, experts etc. Social network analytics community has been successful in introducing and testing such approaches for general purpose social network platforms like Facebook, twitter etc. However, in order to be useful in business context, it is mandatory to integrate underlying business systems, processes and practices with analytics approach. Such integration problem is increasingly recognized as Big Data problem. We argue that Semantic Web technology applied with social network analytics can solve enterprise knowledge management issues while achieving the integration.


Enterprise User Ontology

To enhance human resource management and personalized information acquisition, employee ontology is used to model business concepts and relations between them for enterprises. However, the enterprise knowledge represented in existing ontologies is not sufficient to enable comprehensive semantic search. In this project, we address this problem by taking advantage of employees’ interaction in social media that enterprises adopt. Specially, we mine the knowledge by leveraging the message conversations in enterprise microblogging service, and characterize areas of expertise and interests for individual employees using latent topic descriptors. We build employee ontology by incorporating the discovered knowledge, and illustrate some typical use cases of expert finding, collaboration analysis and team building.


Collective Opinion Mining in Social Media

Social media has created new effective channels of social interaction, which empower people with a variety of shared user-driven content. People can easily share knowledge, ideas and insights within an interactive and collaborative environment. Meanwhile, people can also freely exchange their opinions through thoughts or experience by taking advantage of the open nature of social media. These opinions can be expressed towards different entities including products or services, political campaigns, news events, celebrities, places to travel, movies and etc. For example, when iphone 4S was released and Steve Jobs passed away, there are breaking tweets that contains personal sentiments on Twitter. As social media has enjoyed phenomenal popularity and reached to a broader range of users, it is usually selected as a platform for business marketing as well as political campaign. Customers are now searching for target services, comparing product quality and seek other users’ opinions for reference via social media. For instance, suppose a customer wants to buy the latest released iphone 4S. He or she may post a tweet on Twitter to ask comments about iphone 4S from other users. Credible enterprises are utilizing social media to promote their brands, launch advertisements, analyze sales and target prospective clients. The opinions in social media are hence extremely valuable that can help enterprises finds flaws of products, make improvements and devise adapted business strategy. On the other hand, the customers can greatly benefit from the collective reviews to make purchase decisions by comparison of similar products. Moreover, compared to traditional polls regarding to political issues and consumer confidence, collecting public opinion from freely available text content in social media typically Twitter is a faster and less expensive ways [13]. However, there is usually extremely large volume of opinions scattered in social media. It is difficult for one to digest all of them, especially those which are relevant. This motivates the research in generating overall image that reflects the collective and representative opinions of massive users.

Study of Company Hierarchy

Study of Company Hierarchy
The company hierarchy connects and governs users in a way that shapes the pattern of posting activity, interactions and enacted topics

As popular social media have been adopted by corporations for professional sharing and internal communication, strong ties appear in the networks as employee users communicate with each other on work practices. The company hierarchy connects and governs users in a way that shapes the pattern of posting activity, interactions and enacted topics. In this research project, we quantify the effect of company hierarchy with experiments on a large-scale dataset of enterprise microblogging. Analysis of macroscopic message interaction highlights an pattern that the interactions mainly occur among users at the same hierarchy rank, rather than across hierarchies. We also find a significant upstream message flow that those who are lower-level in hierarchy usually reply to those who are higher-level. We further focus on microscopic patterns at topic level and examine the role of users at various hierarchies in information sharing and expertise exchange. We find that lower-level employees are more influential in providing expertise of technical knowledge, while higher-levels mainly focus on expertise related to decision making and management. Our work has implications in relation building, team collaboration and expertise sharing within a company.


Recommendations in Microblogging Services

Study of Company Hierarchy
Recommendations for each user

Microblogging services such as Twitter and Tencent Weibo have enjoyed drastic popularity in the latest few years. Recommender is essential to those microblogs as a means to find items (users or other information sources such as organizations) that might interest a user to follow. It can greatly improve user experience as well as reduce the risk of information overload might be introduced by irrelevant followees. In this research, we examine some of the most influential factors that user might consider in selecting followees, in the hope of recommending interesting items to match each user’s preferences. We investigate a large scale microblog data extracted from Tencent Weibo and conduct the evaluation of recommendations based on the guideline proposed by the challenge of Track 1 in KDD Cup 2012. Statistical analysis of the log of user actions regarding to recommendations reflect only about 7% acceptance. Experimental results show the popularity of an item is more attractive to users than other features such as the matching of item category, keywords and the influence of user actions and current followees’ acceptance.

Human Data graph in Social Media

Social network services improve social experience by connecting people with shared interest. Similar to real life, seeking good friends is much easier with recommendations in social network services. In this project, we answer a series of questions related to friendship formation in an attempt of improving member matching in social networks. Specially, we investigate whether the users who contribute more annotations are more popular among other users, whether users like to make friends with popular users, how users with different diversity of individual tastes account for friendship formation. A novel approach based on topic modeling is proposed to characterize the interest diversity degree of each user. We then use the diversity features to help predict the friendship between users. The experimental results on three large-scale datasets demonstrate the effectiveness of our method.

Team Members

Current Members

Past Members

Publications

  1. Semantic Social Network Analysis for the Enterprise, Charalampos Chelmis, Hao Wu, Vikram Sorathia, Viktor K. Prasanna, Journal of Computing and Informatics - Special Issue on Computational Intelligence for Business Collaboration, 2014 (To Appear)
  2. Enterprise Knowledge Preservation and Management, Charalampos Chelmis, Vikram Sorathia, Viktor K. Prasanna, Book Chapter, Collaborative Processes and Decision Making in Organizations, IGI Global, 2013 <doi>
  3. Social Link Prediction in Online Social Tagging Systems, Charalampos Chelmis, Viktor K. Prasanna, ACM Transactions on Information Systems, 2013 (To Appear). [pdf]
  4. The Role of Organization Hierarchy in Technology Adoption at the Workplace, Charalampos Chelmis, Viktor K. Prasanna, The 2013 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining, August 2013. [pdf]
  5. Enriching Employee Ontology for Enterprises with Knowledge Discovery from Social Networks, Hao Wu, Charalampos Chelmis, Yinuo Zhang, Vikram Sorathia, Om Patri, Viktor K. Prasanna, The 3rd Workshop on Social Network Analysis in Applications, August 2013. [pdf]
  6. Complex Modelling and Analysis of Workplace Collaboration Data, Charalampos Chelmis, International Conference on Collaboration Technologies and Systems, 2013. [pdf]
  7. An Empirical Analysis of Microblogging Behavior in the Enterprise, Charalampos Chelmis, Viktor K. Prasanna, Social Networking Analysis and Mining, Springer Wien, 2013. <doi>
  8. Exploring Generative Models of Tripartite Graphs for Recommendation in Social Media, Charalampos Chelmis, Viktor K. Prasanna, 4rth ACM International workshop on Modeling social media, May 2013. [pdf]
  9. When Diversity Meets Speciality: Friend Recommendation in Online Social Networks, Hao Wu, Vikram Sorathia and Viktor K. Prasanna, ASE Human Journal, Vol 1, No 1, 2012. [pdf]
  10. Predict Whom One Will Follow: Followee Recommendation in Microblogs, Hao Wu, Vikram Sorathia and Viktor K. Prasanna, ASE International Conference on Social Informatics (SocialInfo), December, 2012. [pdf]
  11. Predicting Communication Intention in Social Networks, Charalampos Chelmis and Viktor K. Prasanna, ASE/IEEE International Conference on Social Computing (SocialCom), September, 2012. [pdf]
  12. Microblogging in the Enterprise: A few comments are in order, Charalampos Chelmis and Viktor K. Prasanna, The 2012 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM), August, 2012. [pdf]
  13. Enterprise Wisdom Captured Socially, Charalampos Chelmis, Vikram Sorathia and Viktor K. Prasanna, The 2012 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM), August, 2012. [pdf]
  14. Social Networking Analysis: A State of the Art and the Effect of Semantics, Charalampos Chelmis and Viktor K. Prasanna, IEEE Third International Conference on Social Computing (SocialCom), October, 2011. [pdf]
Personal tools