RecLab Prize on Overstock.com challenges researchers to advance the state of the art in product recommendations with new privacy-secure cloud environment

San Francisco – May 12, 2011 – Overstock.com (short cut: O.co) and RichRelevance® today unveiled the RecLab Prize on Overstock.com. The Prize provides a cash award totaling up to $1 million to the researcher or research team who can achieve a measurable lift over existing product recommendations in a wide variety of shopping contexts on Overstock.com. The RecLab Prize rewards the highest performing individual or team based on the results they are able to deliver within a defined judging period (up to $1 million for a 10% or greater lift). Complete details about eligibility for registering and competing for the Prize are available at http://overstockreclabprize.com/

RecLab Prize contestants gain immediate access to a high-quality and comprehensive synthetic dataset via RichRelevance’s open-source RecLab project, a highly scalable platform for research code. The RecLab approach enables researchers to develop their algorithms against synthetic data and then test against real data. Top performing algorithms will be exposed to real data and will run live within the RichRelevance cloud environment (as real product recommendations to Overstock.com’s customers). This groundbreaking approach enables researchers to solve a real-world problem with real-world constraints, while never exposing data to an outside system, thereby preserving data security and eliminating privacy concerns.

“We are excited to support the academic research community, bringing state-of-the-art analytics to our business through our partners at RichRelevance,” said Overstock.com CEO Patrick Byrne. “This is a phenomenal opportunity to benefit our customers, who will get early, exclusive access to the most advanced recommendations possible through the participation of top educational and research institutions worldwide.”

Online product recommendations are among the shopping tools most widely used by consumers who need to easily find relevant and enticing products from the millions available online. Overstock.com has worked with RichRelevance since 2009 to present shoppers with dynamic recommendations that grow smarter over time and accurately reflect more than 60 different ways that people shop on the site (by price, by brand, by category). Now Overstock.com is partnering with RichRelevance to open up this real-world business challenge to researchers. The RecLab Prize on Overstock.com breaks rank with previous prizes to present researchers with a complex, multi-dimensional problem facing retailers – not to simply predict how a consumer will rate a product but to effectively pinpoint the most appropriate product array to show shoppers at any point in the research and purchase process. The Prize demands that researchers craft approaches that quickly identify and adapt to contextual clues and maximize every available piece of data.

“The Netflix Prize did a great job of mobilizing the research community around a new and interesting problem. The RecLab Prize on Overstock.com takes the next step by offering researchers the chance to solve a multi-dimensional, real-world problem and see how their best algorithms perform when put in front of live shoppers,” said Darren Vengroff, RichRelevance Chief Scientist and RecLab creator. “The academic community has been clamoring for access to live real-world data and live user interaction to drive their research, and we are giving them more and better access than ever before using a unique approach that absolutely preserves the privacy of shoppers.”

A board of judges, including senior engineers at RichRelevance, Overstock.com, and well-known members of the machine learning community will determine the prize winners. In order to win the $1 million prize, a researcher or team must deliver at least a 10% lift over existing product recommendations on Overstock.com. If no one in the round hits this mark, then the judges will award a pro-rated prize to the team who achieves the highest lift as a percentage of the lift they achieve. For example, if the winning team achieves an 8% lift, it will receive $800,000.

In addition, should the winning team be affiliated with an educational institution, RichRelevance and Overstock.com will grant a separately funded Institution Prize valued at 25% of the winning prize to the educational institution. The RecLab Prize is also open globally to non-commercial teams.