Kusa, W., & Ghafourian, T. (2022). DOSSIER at TREC 2021 Clinical Trials Track. In I. Soboroff & A. Ellis (Eds.), The Thirtieth Text REtrieval Conference (TREC 2021) Proceedings. NIST. http://hdl.handle.net/20.500.12708/153348
The Thirtieth Text REtrieval Conference (TREC 2021) Proceedings
-
Band:
SP 500-335
-
Datum (veröffentlicht):
Feb-2022
-
Veranstaltungsname:
30th TREC 2021
en
Veranstaltungszeitraum:
17-Nov-2021 - 19-Nov-2021
-
Veranstaltungsort:
Gaithersburg, MD, Vereinigte Staaten von Amerika
-
Verlag:
NIST
-
Keywords:
TREC; Clinical Trials matching; TREC Clinical Trials; Information Retrieval
en
Abstract:
This paper describes our experimental setup and results for the Clinical Trials Track at TREC 2021. In particular, we study (i) the effectiveness of post-processing with patients’ metadata and (ii) the novel re-ranking formula using eligibility criteria. We find that the postprocessing improves the model over the baseline run. However, the custom re-ranking negatively impacts average results.
en
Projekttitel:
Domänen-spezifische Systeme für Informationsextraktion und -suche: 860721 (European Commission)
