Three is Better than One: Ensembling Math Information Retrieval Systems

NOVOTNÝ, Vít, Petr SOJKA, Michal ŠTEFÁNIK and Dávid LUPTÁK. Three is Better than One: Ensembling Math Information Retrieval Systems. CEUR Workshop Proceedings. Thessaloniki, Greece: M. Jeusfeld c/o Redaktion Sun SITE, Informatik V, RWTH Aachen., 2020, vol. 2020, No 2696, p. 93-122. ISSN 1613-0073.

Basic information

• Original name: Three is Better than One: Ensembling Math Information Retrieval Systems
• Authors: NOVOTNÝ, Vít (203 Czech Republic, belonging to the institution), Petr SOJKA (203 Czech Republic, guarantor, belonging to the institution), Michal ŠTEFÁNIK (703 Slovakia, belonging to the institution) and Dávid LUPTÁK (703 Slovakia, belonging to the institution).
• Edition: CEUR Workshop Proceedings, Thessaloniki, Greece, M. Jeusfeld c/o Redaktion Sun SITE, Informatik V, RWTH Aachen. 2020, 1613-0073.

Other information

• Original language: English
• Type of outcome: Article in a journal
• Field of Study: 10201 Computer sciences, information science, bioinformatics
• Country of publisher: Greece
• Confidentiality degree: is not subject to a state or trade secret
• WWW: PDF
• RIV identification code: RIV/00216224:14330/20:00116318
• Organization unit: Faculty of Informatics
• Keywords (in Czech): vyhledávání matematiky; odpovědi na otázky; reprezentace matematiky; slovní embedingy; ansámbl
• Keywords in English: math information retrieval; question answering; math representations; word embeddings; ensembling
• Tags: information retrieval, machine learning, math indexing and retrieval, math information retrieval, MIR, SCM, similarity search, soft cosine measure
• Tags: International impact, Reviewed

Abstract

We report on the systems that the Math Information Retrieval group at Masaryk University (MIRMU) prepared for tasks 1 (find answers) and 2 (formula search) of the ARQ Math lab at the CLEF conference. We prototyped three primary MIR systems, proposed several math representations to tackle the lab tasks, and evaluated the proposed systems and representations. We developed a novel algorithm for ensembling information retrieval systems that outperformed all our systems on task 1 and placed ninth out of the 23 competing submissions. Out-of-competition en sembles of all non-baseline primary submissions in the competition made available by the participants placed first on task 1 and third on task 2. Our prototypes will help to understand the challenging problems of answer and formula retrieval in the STEM domain and bring the possibility of accurate math information retrieval one step closer.

Links

• MUNI/A/1076/2019, interní kód MU: Name: Zapojení studentů Fakulty informatiky do mezinárodní vědecké komunity 20 (Acronym: SKOMU)

Investor: Masaryk University, Category A

• MUNI/A/1411/2019, interní kód MU: Name: Aplikovaný výzkum: softwarové architektury kritických infrastruktur, bezpečnost počítačových systémů, zpracování přirozeného jazyka a jazykové inženýrství, vizualizaci velkých dat a rozšířená realita.

Investor: Masaryk University, Category A