UMSI faculty contribute at top venue for Web research
School of Information faculty papers on rumor-detection in social media, contextual disambiguation, and large-scale network embedding will be presented at the International World Wide Web conference, the premier international forum for discussions of progress in research, development and applications of all topics related to the Web. The 24th annual conference takes place in Florence, Italy, May 18-22.
U-M doctoral student Zhe Zhao, UMSI professor Paul Resnick and associate professor Qiaozhu Mei are co-authors of the paper “Enquiring Minds: Early Detection of Rumors in Social Media from Enquiry Posts,” which Zhao will present on Friday, May 22 at 11 a.m.
The researchers have developed a technique that identifies trending rumors in social media by finding clusters of posts whose topic is a disputed factual claim. Signature text phrases that express skepticism, such as “Is that true?,” “Really?,” and “What?,” can help identify rumors online. On a typical day on Twitter, the researchers found, about a third of the top 50 clusters were judged to be rumors, a high enough percentage that human analysts might be willing to examine them more closely.
UMSI associate professor Kevyn Collins-Thompson is co-author of a paper to be presented on Wednesday, May 20, at 2:30 pm on “Contextual Disambiguation for Query Suggestion and Blending.” The paper describes new algorithms that can learn context-sensitive rules for effective query suggestions from large-scale mining of search logs and make reliable decisions about when and how to apply these rules in order to produce better search results for users, especially when their queries are ambiguous.
On Thursday, May 21, at 4:30pm, assistant professor Qiaozhu Mei’s co-authored paper on “LINE: Large-scale information network embedding” will be presented. The authors looked at the problem of embedding very large information networks into low-dimensional vector spaces. They have developed an algorithm that works efficiently on a wide variety of real-world information networks, including language networks, social networks, and citation networks.