To innovate at Helioid we stay on the edge of research and development in search and related technologies. In keeping with this, Helioid co-Founder Peter Lubell-Doughtie will be presenting a poster at the European Conference on Information Retrieval (ECIR) 2012. The poster concerns a paper written by Peter and Katja Hofmann, a PhD student at the University of Amsterdam, titled Learning to Rank from Relevance Feedback for eDiscovery.
eDiscovery involves many relevant documents, a situation not intrinsically handled by main stream search engines, but one that is common place in research searches, be they in the world of academia, legal, finance, or journalism. With so many relevant documents to search through we can substantially improve search performance by learning about user preferences and tailoring search results on a per query and per user basis. Peter and Katja's paper builds a system performs this personalization and shows experimental performance improvements.
You are welcome to view the abstract or read the full paper.