When FastText Pays Attention: Efficient Estimation of Word Representations using Constrained Positional Weighting

NOVOTNÝ, Vít, Michal ŠTEFÁNIK, Eniafe Festus AYETIRAN, Petr SOJKA and Radim ŘEHŮŘEK. When FastText Pays Attention: Efficient Estimation of Word Representations using Constrained Positional Weighting. Journal of Universal Computer Science. New York, USA: J.UCS Consortium, 2022, vol. 28, No 2, p. 181-201. ISSN 0948-695X. Available from: https://dx.doi.org/10.3897/jucs.69619.

Basic information

• Original name: When FastText Pays Attention: Efficient Estimation of Word Representations using Constrained Positional Weighting
• Authors: NOVOTNÝ, Vít (203 Czech Republic, guarantor, belonging to the institution), Michal ŠTEFÁNIK (703 Slovakia, belonging to the institution), Eniafe Festus AYETIRAN (566 Nigeria, belonging to the institution), Petr SOJKA (203 Czech Republic, belonging to the institution) and Radim ŘEHŮŘEK (203 Czech Republic).
• Edition: Journal of Universal Computer Science, New York, USA, J.UCS Consortium, 2022, 0948-695X.

Other information

• Original language: English
• Type of outcome: Article in a journal
• Field of Study: 10201 Computer sciences, information science, bioinformatics
• Country of publisher: Czech Republic
• Confidentiality degree: is not subject to a state or trade secret
• WWW: preprint DOI
• Impact factor: Impact factor: 1.000
• RIV identification code: RIV/00216224:14330/22:00124923
• Organization unit: Faculty of Informatics
• Doi: http://dx.doi.org/10.3897/jucs.69619
• UT WoS: 000767374300005
• Keywords in English: Word embeddings; fastText; attention
• Tags: attention, language modeling, Positional embeddings, word embeddings
• Tags: International impact, Reviewed

Abstract

In 2018, Mikolov et al. introduced the positional language model, which has characteristics of attention-based neural machine translation models and which achieved state-of-the-art performance on the intrinsic word analogy task. However, the positional model is not practically fast and it has never been evaluated on qualitative criteria or extrinsic tasks. We propose a constrained positional model, which adapts the sparse attention mechanism from neural machine translation to improve the speed of the positional model. We evaluate the positional and constrained positional models on three novel qualitative criteria and on language modeling. We show that the positional and constrained positional models contain interpretable information about the grammatical properties of words and outperform other shallow models on language modeling. We also show that our constrained model outperforms the positional model on language modeling and trains twice as fast.

Links

• MUNI/A/1195/2021, interní kód MU: Name: Aplikovaný výzkum v oblastech vyhledávání, analýz a vizualizací rozsáhlých dat, zpracování přirozeného jazyka a aplikované umělé inteligence

Investor: Masaryk University