Introducing LETOR 4.0 Datasets

摘要: LETOR is a package of benchmark data sets for research on LEarning TO Rank, which contains standard features, relevance judgments, partitioning, evaluation tools, and several baselines. Version 1.0 was released in April 2007. 2.0 Dec. 3.0 2008. This version, 4.0, July 2009. Very different from previous versions (V3.0 an update based V2.0 V1.0), LETOR4.0 totally new release. It uses the Gov2 web page collection (~25M pages) two query Million Query track TREC 2007 We call MQ2007 MQ2008 short. There are about 1700 queries with labeled documents 800 documents. If you have any questions or suggestions datasets, please kindly email us (letor@microsoft.com). Our goal to make dataset reliable useful community.