|Tony Fader (BS Math 2007) won a prestigious NSF Graduate Research Fellowship while conducting research under Prof. Drago Radev. This research spanned two different projects:|
This project aims to develop a method to find influential speakers (or mavens) from the transcript of a debate. The method, called MavenRank, represents speeches and their textual similarity as a network and ranks speakers based on the centrality of their speeches. When applied to the US Congressional Record, which is a transcript of debates and speeches in the US Senate and House of Representatives, it was found that the rank of a speaker in a congressional committee is correlated with that speaker's MavenRank score in a related topic. Work is now underway to develop a dynamic version of MavenRank that identifies influential speakers within a specific time period.
In the fields of biology and medicine, researchers are often overwhelmed by the large volume of articles published, which makes keeping up to date with a specific topic difficult. Radev and Fader developed a system called GIN (Gene Interaction Network), which aims to help researchers find what they are looking for by providing article summaries and access molecule interaction networks. Each interaction is automatically extracted from the text, so users can immediately find more information about a given interaction or view all of the interactions described in an article. The system also provides network statistics about the molecules in the interaction network, which describe the centrality of the molecules and connectedness of their neighborhoods.
Tony will be headed to U. Washington to pursue his PhD in Computer Science.