Dan Pelleg

Haifa, Israel

I am a data scientist at Yahoo research Haifa, working on new avenues for Yahoo Answers. I look at data a lot, to see at what engages users, what makes them leave, and how to improve retention. The big challenge is to improve experience and return for both users and Yahoo. Along this track I work on new-user retention schemes, extract monetization-boosting data, and try to characterize the effect of mobile device use on both user interaction and language. I lead a research team which conducts research in community question-answering, analyzing both users' behavior and answers content, while conducting advanced research in information retrieval, natural language processing, web data mining, applied machine learning and social media research. The team tightly collaborates with the Yahoo! Answers engineering team, providing key insights, and delivering new features. Previously, I worked for the Analytics department at IBM Haifa, leading the green analytics activity, and applying machine learning to areas such as hardware verification, CRM, energy demand forecast, virtual machine and operating system monitoring, storage products, software engineering, and water management. Even before that, I worked (in different phases) on bioinformatics, web search, computational astrostatistics, clustering, and accelerated data-mining algorithms. I got my computer science B.Sc. (1995) and M.Sc. (1998) from the Technion, Israel, and Ph.D. (2004) from Carnegie-Mellon university. I filed several patents (14 issued, last time I counted), and in 2009 was appointed IBM Master Inventor. In 2008, I received an IBM Research Technical Accomplishment award for my work in virtual machine availability for IBM Systems Director. I also contributed to the Alphaworks release of the Parallel Machine Learning framework (PML). Over time, I released some of my data mining code as open source. I run the Israeli machine learning site. I regularly serve on committees for conferences such as WWW and WSDM. I bake bread, grow my own sourdough, and pickle veggies.

Mail, Metrics and User Engagement, Natural Language Processing, Search