Home

Awesome

Kyoto University Text Corpus

Overview

This is a text corpus that is manually annotated with various linguistic information. It consists of approximately 40,000 sentences from Mainichi newspaper in 1995 with morphological and syntactic annotations. Out of these sentences, approximately 20,000 sentences are annotated with predicate-argument structures including zero anaphora and coreferences.

Notes

This repository does not include original sentences but include only annotation information. To recover the complete annotated corpus, it is necessary to obtain the Mainichi 1995 CD-ROM. The information of this CD-ROM is available at https://www.nichigai.co.jp/sales/corpus.html

Notes on annotation guidelines

The annotation guidelines for this corpus are written in the manuals found in "doc" directory. The guidelines for morphology and dependencies are described in syn_guideline.pdf, and those for predicate-argument structures and coreferences are described in rel_guideline.pdf.

Distributed files

Conversion to the complete annotated corpus

The distributed files contain only annotation information. To recover the complete annotated corpus, it is necessary to obtain the Mainichi 1995 CD-ROM. The conversion process is the following.

  1. Mount the Mainichi 1995 CD-ROM. (Here, suppose that it is mounted on /mnt/cdrom.)
  2. Execute ./auto_conv -d /mnt/cdrom.

As a result of this process, *.knp files are generated under dat/syn/ and dat/rel/.

The corpus annotated with morphology and dependencies is stored under dat/syn/, and the corpus annotated with morphology, dependencies, predicate-argument structures, and coreferences is stored under dat/rel/. The encoding of the files is UTF-8.

The above conversion process works on UNIX systems. If you have a problem on other systems, please let us know.

Format of the corpus

The annotated corpus (after conversion) has the following format.

# S-ID:950101001-001
* 2D
+ 3D
太郎 たろう 太郎 名詞 6 人名 5 * 0 * 0
は は は 助詞 9 副助詞 2 * 0 * 0
* 2D
+ 2D
京都 きょうと 京都 名詞 6 地名 4 * 0 * 0
+ 3D
大学 だいがく 大学 名詞 6 普通名詞 1 * 0 * 0
に に に 助詞 9 格助詞 1 * 0 * 0
* -1D
+ -1D <rel type="ガ" target="太郎" sid="w201106-0000010001-1" id="0"/><rel type="ニ" target="大学" sid="w201106-0000010001-1" id="2"/>
行った いった 行く 動詞 2 * 0 子音動詞カ行促音便形 3 タ形 10
EOS

The first line represents the ID of this sentence. In the subsequent lines, the lines starting with "*" denote "bunsetsu," the lines starting with "+" denote basic phrases, and the other lines denote morphemes.

The line of morphemes is the same as the output of the morphological analyzers, JUMAN and Juman++. This information includes surface string, reading, lemma, part of speech (POS), fine-grained POS, conjugate type, and conjugate form. "*" means that its field is not available.

The line starting with "*" represents "bunsetsu," which is a conventional unit for dependency in Japanese. "Bunsetsu" consists of one or more content words and zero or more function words. In this line, the first numeral means the ID of its depending head. The subsequent alphabet denotes the type of dependency relation, i.e., "D" (normal dependency), "P" (coordination dependency), "I" (incomplete coordination dependency), and "A" (appositive dependency).

The line starting with "+" represents a basic phrase, which is a unit to which various relations are annotated. A basic phrase consists of one content word and zero or more function words. Therefore, it is equivalent to a bunsetsu or a part of a bunsetsu. In this line, the first numeral means the ID of its depending head. The subsequent alphabet is defined in the same way as bunsetsu. The remaining part of this line includes the annotations of named entity and various relations.

Annotations of various relations are given in <rel> tags (only in dat/rel/*.knp files). <rel> has the following four attributes: type, target, sid, and id, which mean the name of a relation, the string of the counterpart, the sentence ID of the counterpart, and the basic phrase ID of the counterpart, respectively. If a basic phrase has multiple tags of the same type, a "mode" attribute is also assigned, which has one of "AND," "OR," and "？." The details of these attributes are described in the annotation guidelines (rel_guideline.pdf).

Note that this format is slightly different from Kyoto University Text Corpus 4.0, but it is the same as KNP and Kyoto University Web Document Leads Corpus (KWDLC).

Terms of use

Disclaimer: The developers, Kurohashi Laboratory of Kyoto University and Kawahara Laboratory of Waseda University, are not liable for any damages incurred from the use of the Kyoto University Text Corpus.

Commercial Use: If you intend to use the corpus for commercial purposes, kindly include a statement in the documentation or other materials of your commercial system acknowledging the utilization of the Kyoto University Text Corpus.

For inquiries related to the commercial use of the Mainichi Shimbun original texts, we recommend contacting The Mainichi Newspapers Co. or Nichigai Associates, Inc.

Reference

Contact

If you have any questions or problems about this corpus, please send an email to nl-resource at nlp.ist.i.kyoto-u.ac.jp.