Abstract
This paper describes our entry into the TREC 2011 Microblog track. We submitted two runs in this year's track, both in real-time fashion and without any external resources. The runs were generated through a three-step procedure, including scoring, threshold selection and re-ranking. The evaluation results of our submitted runs significantly outperform the disjunctive baseline run. We conducted additional runs to show our score decay and threshold selection strategies to be exceptionally effective.
| Original language | English |
|---|---|
| Journal | NIST Special Publication |
| State | Published - 2011 |
| Event | 20th Text REtrieval Conference, TREC 2011 - Gaithersburg, MD, United States Duration: 15 Nov 2011 → 18 Nov 2011 |
Fingerprint
Dive into the research topics of 'HIT-LTRC at TREC 2011 microblog track'. Together they form a unique fingerprint.Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver