----- Forwarded Message ----
From: Tie-Yan Liu <tyliu@microsoft.com>
Sent: Tuesday, January 1, 2008 3:48:43 PM
Subject: [Dbworld] Major Update of LETOR, a Benchmark Dataset for Learning to Rank
Subject: Major Update of LETOR, a Benchmark Dataset for Learning to Rank
Author: Tie-Yan Liu (
tyliu@microsoft.com).
Body:
Microsoft Research Asia released a benchmark dataset for research on learning to rank, named LETOR, in March 2007. This dataset contains standard features, relevance judgments, data partitioning, evaluation tools, and several baselines, for the OHSUMED data collection and the '.gov' data collection. The detailed information about LETOR can be found at
http://research.microsoft.com/users/tyliu/LETOR/.
In the past year, LETOR has been used in many research papers on learning to rank. At the same time, we have received valuable feedback from many people, including bug reports, feasibility studies of the tools, and other suggestions. Based on the feedback, we have made a major update to the LETOR dataset as described below.
1) Bugs fixed in the baseline experimental results and the evaluation tools.
2) More baseline experimental results provided, including ListNet (ICML 2007), AdaRank (SIGIR 2007), FRank (SIGIR 2007) and MHR (SIGIR 2007).
3) Evaluation tools updated, including a tool which receives ranked list directly instead of ranking scores as input, and a significant test tool.
3) A channel opened for accepting new baselines from other researchers.
4) A discussion board created for researchers to exchange their ideas on learning to rank and the LETOR dataset.
5) More information collected about active research groups, papers and other resources on learning to rank.
Accompanying the new update, we have created a new website for LETOR, and made it online on the first day of 2008. You can visit the new website by following this hyperlink,
http://research.microsoft.com/users/LETOR/, or find the way of transiting to it at the previous website of LETOR.
In this New Year's Day, we would like to express our sincere thanks to you, for your helps and suggestions on LETOR in the past year. We hope that we can continue hearing from you in the year of 2008. Please feel free to send an email to
letor@microsoft.com if you want to contact us.
Best Regards and Happy New Year!
Tie-Yan Liu
Microsoft Research Asia
http://research.microsoft.com/users/tyliu/
感谢新老用户对希网网络支持和关爱，以下杂志是一些精彩的电子杂志。
没有评论:
发表评论