I wrote a rudimentary piece of software to load and analyse Microsoft LETOR 3.0 datasets (http://research.microsoft.com/en-us/um/beijing/projects/letor/).
- Can load/save multiple LETOR files (tested with the OSHUMED and GOV datasets)
- Can export to different formats i.e. MATLAB
- Can Normalize the data
- Gives query/document statistics
- Allows you to select which query/documents and features to use as test/train
- It runs the SVMRank program by Thorsten Joachims with configurable attributes
- Visualize the relative weights of the selected features
- P@k score (T. Qin, T.Y. Liu, J. Xu, and H. Li. How to Make LETOR More Useful and Reliable)
- Requires .NET 4.0 (Windows)
Screenshots
|
 Updating...
Dr. Tim Scarfe, 4 Mar 2013, 12:57
|