English

Unsupervised Tokenization Learning

Computation and Language 2022-12-16 v4 Artificial Intelligence Symbolic Computation

Abstract

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.

Keywords

Cite

@article{arxiv.2205.11443,
  title  = {Unsupervised Tokenization Learning},
  author = {Anton Kolonin and Vignav Ramesh},
  journal= {arXiv preprint arXiv:2205.11443},
  year   = {2022}
}

Comments

16 pages, 9 figures; Paper accepted to the EMNLP 2022 conference

R2 v1 2026-06-24T11:25:55.523Z