Результаты исследований: Материалы конференций › материалы › Рецензирование
Unsupervised Tokenization Learning. / Kolonin, Anton; Ramesh, Vignav.
2022. Работа представлена на 2022 Conference on Empirical Methods in Natural Language Processing, Абу-Даби, Объединенные Арабские Эмираты.Результаты исследований: Материалы конференций › материалы › Рецензирование
}
TY - CONF
T1 - Unsupervised Tokenization Learning
AU - Kolonin, Anton
AU - Ramesh, Vignav
PY - 2022
Y1 - 2022
N2 - In the presented study, we discover that the so-called “transition freedom” metric appears superior for unsupervised tokenization purposes in comparison to statistical metrics such as mutual information and conditional probability, providing F-measure scores in range from 0.71 to 1.0 across explored multilingual corpora. We find that different languages require different offshoots of that metric (such as derivative, variance, and “peak values”) for successful tokenization. Larger training corpora do not necessarily result in better tokenization quality, while compressing the models by eliminating statistically weak evidence tends to improve performance. The proposed unsupervised tokenization technique provides quality better than or comparable to lexicon-based ones, depending on the language.
AB - In the presented study, we discover that the so-called “transition freedom” metric appears superior for unsupervised tokenization purposes in comparison to statistical metrics such as mutual information and conditional probability, providing F-measure scores in range from 0.71 to 1.0 across explored multilingual corpora. We find that different languages require different offshoots of that metric (such as derivative, variance, and “peak values”) for successful tokenization. Larger training corpora do not necessarily result in better tokenization quality, while compressing the models by eliminating statistically weak evidence tends to improve performance. The proposed unsupervised tokenization technique provides quality better than or comparable to lexicon-based ones, depending on the language.
UR - https://www.scopus.com/record/display.uri?eid=2-s2.0-85149438240&origin=inward&txGid=fc05dad2329d756230392a5e18aba688
UR - https://www.mendeley.com/catalogue/1872a78e-654b-330c-9b2b-518308d21af9/
U2 - 10.18653/v1/2022.emnlp-main.239
DO - 10.18653/v1/2022.emnlp-main.239
M3 - Paper
T2 - 2022 Conference on Empirical Methods in Natural Language Processing
Y2 - 7 December 2022 through 11 December 2023
ER -
ID: 55718343