English

Mostly-Unsupervised Statistical Segmentation of Japanese Kanji Sequences

Computation and Language 2007-05-23 v1

Abstract

Given the lack of word delimiters in written Japanese, word segmentation is generally considered a crucial first step in processing Japanese texts. Typical Japanese segmentation algorithms rely either on a lexicon and syntactic analysis or on pre-segmented data; but these are labor-intensive, and the lexico-syntactic techniques are vulnerable to the unknown word problem. In contrast, we introduce a novel, more robust statistical method utilizing unsegmented training data. Despite its simplicity, the algorithm yields performance on long kanji sequences comparable to and sometimes surpassing that of state-of-the-art morphological analyzers over a variety of error metrics. The algorithm also outperforms another mostly-unsupervised statistical algorithm previously proposed for Chinese. Additionally, we present a two-level annotation scheme for Japanese to incorporate multiple segmentation granularities, and introduce two novel evaluation metrics, both based on the notion of a compatible bracket, that can account for multiple granularities simultaneously.

Keywords

Cite

@article{arxiv.cs/0205009,
  title  = {Mostly-Unsupervised Statistical Segmentation of Japanese Kanji Sequences},
  author = {Rie Kubota Ando and Lillian Lee},
  journal= {arXiv preprint arXiv:cs/0205009},
  year   = {2007}
}

Comments

22 pages. To appear in Natural Language Engineering

R2 v1 2026-07-22T12:19:48.779Z