LexisNexis Learning to Rank Case Study
Doug Rosenoff • Back to Haystack 2018
Using a large sample of user searches and the documents that were engaged from those searches, LexisNexis set out to predict and re-rank the top answers so that engaged documents would appear closer to the top of the Search Engine Result Page (SERP). Lexis applied a three phased approach to collect and compile LETOR data to drive a Java LambdaMART implementation that re-ranked the top answers in the SERP. Statistical, Human Relevance and Search A/B testing methods were used to verify and analyze both the human and engaged Discounted Cumulative Gain score results.
Doug Rosenoff is currently Director of Search for the Global Product Management team of Lexis Nexis. He is primarily involved in enhancing and developing search relevance algorithms and metrics, and is the lead for several teams working in those areas. Prior to working at Lexis, he spent 22 years working for Thomson Reuters and West Publishing in a variety of technical and product management positions. Before turning to lexical and semantic search problems, he spent thirteen years in Geophysical Research at Phillips Petroleum Company. A graduate of the Colorado School of Mines, he holds a number of US and International patents in linking and search algorithms.