Tracking Influence and Opinions in Social Media

Social Media such as blogs, wikis, formus and user-generated content sites like flickr, delicious and youtube have become both a source of information and entertainment. The size of audience that these sites currently yield is already rivaling traditional main stream media sources like television, newspapers and magazines. Blogs, especially, have been reported to play a notable role in influencing the buying patterns of consumers. Often a buyer looks for opinions, user experiences and reviews or products on forums, blogs and wikis before purchasing a product.

In this talk we present some ideas on modeling influence in social media, based on our experience with this year's TREC opinion extraction task. In particular, we describe techniques to automatically identify key individuals who play an important role in propagating information. We propose extensions to incorporate topical and readership-based measures of influence. An important component in understanding influence is to detect sentiment and opinions. Aggregated opinions over many users is a predictor for an interesting trend in a community. Sufficient adoption of this trend could lead to a 'tipping point' and consequently influencing the rest of the community. By considering influence as a temporal phenomenon, we propose to find the 'buzz generators' and 'early adopters' in such communities.

Detecting influence and understanding its role in how people perceive and adopt a product or service provides a powerful tool for marketing, advertising and business intelligence. This requires new algorithms that build on social network analysis, community detection and opinion extraction.