TREC NeuCLIR Technical Documents

[Sean MacAvaney]

Hi TREC NeuCLIRs!

In case you have not heard, this year's new task focuses on technical document retrieval. The deadline is August 14, so you still have a few weeks to participate.

The task is to retrieve technical abstracts written in Chinese using either English (cross-language) or Chinese (monolingual) queries. The corpus is called CSL and is available here. The evaluation topics are found on the TREC website. We also have baseline results (e.g., if you want to re-rank) posted on the Slack channel.

If you already have a Chinese cross-language or monolingual retrieval system (either from this year or last year), we encourage you to try it out in the technical setting. It shouldn't be very challenging or time-consuming to run (the corpus is smaller than the news corpus, with ~400k documents), but we expect the technical domain to expose new challenges in CLIR.

If you have any questions, feel free to reach out!

Best,

NeuCLIR Organizers