Ensembling Math Information Retrieval Systems: MIRMU and MSM at ARQMath 2021

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216224%3A14330%2F21%3A00121922" target="_blank" >RIV/00216224:14330/21:00121922 - isvavai.cz</a>

Výsledek na webu

<a href="http://ceur-ws.org/Vol-2936/paper-06.pdf" target="_blank" >http://ceur-ws.org/Vol-2936/paper-06.pdf</a>

DOI - Digital Object Identifier

—

Jazyk výsledku

angličtina

Název v původním jazyce

Ensembling Math Information Retrieval Systems: MIRMU and MSM at ARQMath 2021

Popis výsledku v původním jazyce

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. Using the results of the ten "weak" systems, we ensembled them into committees and let them vote to provide question answers. We evaluated the proposed systems and ensembles, considering their diversity, hyperparameters, and representations used, and classified their approaches. We evidenced the diversity of all systems and evaluated three voting algorithms to collect and rank the answers. Ensembling techniques typically 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 ensemble evaluation framework sets a new direction in MIR research. Finally, we formulate ten commandments for future work in the area.

Název v anglickém jazyce

Ensembling Math Information Retrieval Systems: MIRMU and MSM at ARQMath 2021

Popis výsledku anglicky

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. Using the results of the ten "weak" systems, we ensembled them into committees and let them vote to provide question answers. We evaluated the proposed systems and ensembles, considering their diversity, hyperparameters, and representations used, and classified their approaches. We evidenced the diversity of all systems and evaluated three voting algorithms to collect and rank the answers. Ensembling techniques typically 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 ensemble evaluation framework sets a new direction in MIR research. Finally, we formulate ten commandments for future work in the area.

Druh

D - Stať ve sborníku

CEP obor

—

OECD FORD obor

10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)

Projekt

—

Návaznosti

S - Specificky vyzkum na vysokych skolach

Rok uplatnění

2021

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název statě ve sborníku

CEUR Workshop Proceedings, Volume 2936: 2021 Working Notes of CLEF - Conference and Labs of the Evaluation Forum, CLEF-WN 2021

ISBN

—

ISSN

1613-0073

e-ISSN

—

Počet stran výsledku

25

Strana od-do

82-106

Název nakladatele

CEUR-WS

Místo vydání

Bucharest, Romania

Místo konání akce

Bucharest, Romania

Datum konání akce

1. 1. 2021

Typ akce podle státní příslušnosti

CST - Celostátní akce

Kód UT WoS článku

—