Other formats:
BibTeX
LaTeX
RIS
@inproceedings{1793382,
author = {Novotný, Vít and Štefánik, Michal and Lupták, Dávid and Geletka, Martin and Zelina, Petr and Sojka, Petr},
address = {Bucharest, Romania},
booktitle = {CEUR Workshop Proceedings},
editor = {Guglielmo Faggioli},
keywords = {math information retrieval; question answering; math representations; word embeddings; ensembling},
howpublished = {elektronická verze "online"},
language = {eng},
location = {Bucharest, Romania},
pages = {82-106},
publisher = {M. Jeusfeld c/o Redaktion Sun SITE, Informatik V, RWTH Aachen},
title = {Ensembling Ten Math Information Retrieval Systems: MIRMU and MSM at ARQMath 2021},
url = {http://ceur-ws.org/Vol-2936/paper-06.pdf},
year = {2021}
}
TY - JOUR
ID - 1793382
AU - Novotný, Vít - Štefánik, Michal - Lupták, Dávid - Geletka, Martin - Zelina, Petr - Sojka, Petr
PY - 2021
TI - Ensembling Ten Math Information Retrieval Systems: MIRMU and MSM at ARQMath 2021
PB - M. Jeusfeld c/o Redaktion Sun SITE, Informatik V, RWTH Aachen
CY - Bucharest, Romania
KW - math information retrieval
KW - question answering
KW - math representations
KW - word embeddings
KW - ensembling
UR - http://ceur-ws.org/Vol-2936/paper-06.pdf
N2 - We report on the systems that the Math Information Retrieval group at Masaryk University (MIRMU) and the team of Faculty of Informatics students (MSM) prepared for task~1 (find answers) of the ARQMath lab at the CLEF conference. We have prototyped ten math-aware information retrieval (MIR) systems for the main question-answering task. We ensembled the results of the ten ``weak'' individual systems into committees and let them vote to provide answers to questions. We evaluated the proposed individual systems and ensembles, considering their diversity, hyperparameters, and representations used, and classified their approaches. We have shown the diversity of all systems and evaluated four voting algorithms to collect and rank the answers. Ensembling techniques consistently outperformed the base systems and showed the power of voting of diverse systems. Our prototypes help to understand the challenging problems of question-answering in the STEM domain and our novel reproducible evaluation framework sets a new direction in MIR research. Finally, we formulate ten commandments for future work in the area.
ER -
NOVOTNÝ, Vít, Michal ŠTEFÁNIK, Dávid LUPTÁK, Martin GELETKA, Petr ZELINA and Petr SOJKA. Ensembling Ten Math Information Retrieval Systems: MIRMU and MSM at ARQMath 2021. Online. In Guglielmo Faggioli. \textit{CEUR Workshop Proceedings}. Bucharest, Romania: M. Jeusfeld c/o Redaktion Sun SITE, Informatik V, RWTH Aachen, 2021, p.~82-106. ISSN~1613-0073.
|
