ir_datasets icon indicating copy to clipboard operation
ir_datasets copied to clipboard

high recall retrieval datasets

Open seanmacavaney opened this issue 4 years ago • 1 comments

For testing e.g., total recall systems.

Datasets include RCV1 and JebBush.

The document collections are not publicly available, but we can provide instructions on how to get copies, just like for trec-robust04 and the TREC multilingual datasets.

@eugene-yang is an expert on these. Maybe he can help.

seanmacavaney avatar Nov 13 '20 03:11 seanmacavaney

Sure thing!

eugene-yang avatar Nov 13 '20 03:11 eugene-yang