User Expertise Discovery in Online Domains

Опубликовано 12 августа 2016, 0:12

Online communities such as question answering, microblogs, blogs, social networking sites, provide an excellent platform for their users to generate content. User generated content is the main reason why some of the most popular sites on the Web sustain - Facebook, Wikipedia, Yahoo Answers, etc. The user-centric paradigm of content creation taps on the wisdom of crowd. This idea has turned out to be quite useful in creating vast knowledge sources on the web. But at the same time, these sources are vulnerable to the whims of their users', especially malicious users, vandals, spammers, advertisers. Discovering the expertise and authoritativeness of the content creators can help protect and may be improve these knowledge sources. In this talk, I will describe our work on expertise discovery on large scale datasets from microblogs and question answering systems. We posit that expertise discovery can provide a deeper understanding of how users' evolve, how they influence other community members, and what measures can be taken to improve their participation in the community. I will also discuss the issue of how soon can we identify the expertise of a user. I will illustrate some potential applications of our work in search engines, user retention mechanisms, and recommendation systems.