Jeff Pasternack

I am a recent Ph.D. (now interim post-doc) in computer science at the University of Illinois, Urbana-Champaign, working in the CCG group with Dan Roth.

I research topics in machine learning, natural language processing, text mining, and information retrieval.  Recent work includes papers on transliteration, information extraction (maximum subsequence segmentation), and constrained structured learning.  My main dissertation research concerns determining and measuring the trustworthiness of information, identifying which statements and claims are true (and which are not) in large corpora with diverse authorship.  This is complicated by both the subjective nature of truth (a philosophical point with deep practical implications) and the need to efficiently process web-scale data; a key challenge has thus been finding trust models that can tractably incorporate the user's background knowledge and prior beliefs.