Mostly-Unsupervised Statistical Segmentation of Japanese Kanji Sequences

dc.creatorAndo, Rie Kubota
dc.creatorLee, Lillian
dc.date2002-05-10
dc.date.accessioned2026-07-07T03:18:23Z
dc.date.available2026-07-07T03:18:23Z
dc.descriptionGiven 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.
dc.description22 pages. To appear in Natural Language Engineering
dc.identifierhttps://arxiv.org/abs/cs/0205009
dc.identifierhttp://arxiv.org/abs/cs/0205009
dc.identifierNatural Language Engineering 9 (2), pp. 127--149, 2003
dc.identifierdoi:10.1017/S1351324902002954
dc.identifier.urihttp://salesiana.dossiersoluciones.com/handle/123456789/31091
dc.subjectComputation and Language
dc.subjectI.2.7
dc.titleMostly-Unsupervised Statistical Segmentation of Japanese Kanji Sequences
dc.typetext

Files

Collections