LETOR: Learning to Rank for Information Retrieval - Microsoft Research
learning-to-rankinformation-retrievalbenchmarkmicrosoft-research
Abstraction: Microsoft Research benchmark suite for learning-to-rank IR research
Key points:
- LETOR provides standard features, relevance judgments, data partitioning, evaluation tools, and baselines
- Version 1.0 released April 2007; 4.0 released July 2009 as a complete redesign
- LETOR 4.0 uses the Gov2 web collection (~25M pages) with MQ2007 (~1700 queries) and MQ2008 (~800 queries) from TREC Million Query tracks
- LETOR 3.0 cited as: Qin, Liu, Xu, Li, Information Retrieval Journal 2010
- LETOR 4.0 cited as: Qin and Liu, arXiv:1306.2597
Connections: Microsoft Research · Letor · Learning To Rank · Information Retrieval
Source: http://research.microsoft.com/en-us/um/beijing/projects/letor//