Victor Lavrenko

I'm a Lecturer in Informatics at the University of Edinburgh (that's British speak for assistant professor of computer science). I work on developing better algorithms for search engines, with a particular focus on interaction, multimedia and scalability. I also supervise a number of PhD / MSc students and teach popular courses on search engines and applied machine learning. I am the author and maintainer of yari:mtx -- a swiss-army knife for large datasets.

Prior to Edinburgh, I directed AEnalytics -- a consulting firm in St.Petersburg, Russia. We built one of the world's first news-based arbitrage systems for Credit Suisse. We monitored half-a-million news stories per day, looking for events that may affect stock prices, and continuously re-balancing the portfolio. The system was based on my earlier work on AEnalyst.

Before that, I was a PhD student, and later a post-doc at UMass Amherst, working with Bruce Croft, James Allan, and many others. I worked on a number of projects, but my primary contributions came in three areas: (1) I invented relevance models: a pretty-good formula for query expansion, which to this day wins TREC competitions; (2) I developed a family of highly-cited methods for predicting tags for images; and (3) I built the UMass Topic Detection and Tracking system, which was among the best in DARPA's TDT competitions.