Publications

• Inducing Gazetteers for Named Entity Recognition by Large-scale Clustering of Dependency Relations. Jun'ichi Kazama and Kentaro Torisawa. (to appear In ACL-08: HLT) (pdf) (software)
  

• Webの資源から構築した複数の固有表現辞書を用いた日本語固有表現認識 風間 淳一  鳥澤 健太郎. 言語処理学会第14回年次大会. March, 2008, Tokyo, Japan. (in Japanese) (pdf)
  

• A New Perceptron Algorithm for Sequence Labeling with Non-local Features. Jun'ichi Kazama and Kentaro Torisawa. In Proceedings of the 2007 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning (EMNLP-CoNLL 2007) (Poster), pp. 315-324, June, 2007, Prague, Czech Republic. (pdf)
  

• Exploiting Wikipedia as External Knowledge for Named Entity Recognition. Jun'ichi Kazama and Kentaro Torisawa. In Proceedings of the 2007 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning (EMNLP-CoNLL 2007), pp. 698-707, June, 2007, Prague, Czech Republic. (pdf) (Wikipedia category info for CoNLL 2003 dataset)
  

• 大域的素性を用いたタグ付けのためのパーセプトロン学習 風間 淳一  鳥澤 健太郎. 言語処理学会第13回年次大会. March, 2007, Seta, Shiga, Japan. (in Japanese) (pdf)
  

• Semantic Role Recognition using Kernels on Weighted Marked Ordered Labeled Trees. Jun'ichi Kazama and Kentaro Torisawa. In Proceedings of the Tenth Conference on Computational Natural Language Learning (CoNLL-X), pp. 53-60. June, 2006, New York City, USA. (pdf)
  

• 「木構造カーネルの高速化とノード関係分類への応用」 風間 淳一  鳥澤 健太郎. 言語処理学会第12回年次大会. March, 2006, Yokohama, Japan. (in Japanese) (pdf)
  

• 「属性語のWeb文書からの自動獲得と人手評価のための基準」 徳永 耕亮  風間 淳一  鳥澤 健太郎. 自然言語処理 13(4), pp. 49-67, October, 2006. (in Japanese) (pdf)
  

• Speeding up Training with Tree Kernels for Node Relation Labeling.  Jun'ichi Kazama and Kentaro Torisawa. In Proceedings of the Human Language Technology Conference and Conference on Empirical Methods in Natural Language Processing (HLT/EMNLP 2005), pp. 137-144. October, 2005, Vancouver, Canada. (pdf). software
  
  • Errata: In the 2nd page, we forgot to mention that the conditions for Collins kernel include "and the number of nodes >= 2."
  • Errata: By mistake, we wrote in the experiment section that we used the outputs of Collins' parser, but the fact was that the outputs of Charniak's parser were used.

• Maximum Entropy Models with Inequality Constraints: A case study on text categorization.  Jun'ichi Kazama and Jun'ichi Tsujii. Machine Learning (Special Issue on Learning in Speech and Language Technologies), 60(1-3), pp. 159-194. September, 2005. abstract (publisher site)(pdf)

• Automatic Discovery of Attribute Words from Web Documents.  Kosuke Tokunaga, Jun'ichi Kazama , and Kentaro Torisawa. In LNAI 3651, Natural Language Processing - IJCNLP 2005, Dale et al. (Eds.), pp. 106-118. October, 2005, Jeju, Korea. (pdf)
  
  • Errata:In the 3rd page "Here, the nouns are the words tagged as "proper noun", "sahen noun" (...), "location", or "unknown" (e.g., words written in katakana). => "Here, the nouns are the words tagged as "common noun", "sahen noun" (...), "location", or "unknown" (words written in katakana or alphabet). (revised pdf)

• 「HTML文書からの属性語の自動抽出」 徳永 耕亮  風間 淳一  鳥澤 健太郎. 言語処理学会第11回年次大会. March, 2005, Takamatsu, Japan. (in Japanese)

• 「教師なし隠れマルコフモデルを利用した最大エントロピータグ付けモデル」  風間 淳一  宮尾 祐介   辻井 潤一. 自然言語処理 11(4), pp. 3-23. October, 2003. (in Japanese) (pdf)
  

• Improving Maximum Entropy Natural Language Processing by Uncertainty-aware Extensions and Unsupervised Learning.   Jun'ichi Kazama. Ph.D. Dissertation. University of Tokyo. March, 2004. cover (pdf) body (pdf)
• Warning: There are several heavy figures, which may slow down viewers such as Acrobat reader (Not freezing, wait  until the figures are displayed)
  

• Evaluation and Extension of Maximum Entropy Models with Inequality Constraints.  Jun'ichi Kazama and Jun'ichi Tsujii. In Proceedings of the 2003 Empirical Methods in Natural Language Processing (EMNLP 2003), pp. 137-144. July, 2003, Sapporo, Japan. (ps)(pdf)
• Updates:  we forgot to mention that a document vector was normalized so that the sum of all elements of the vector is 1 (1-norm).
• Errata: we forgot to put "-" (minus) sign in the definition of H(p) (entropy). (revised ps) (revised pdf)
  

• Extracting Biomedical Ontology from Textbooks and Article Abstracts.  Yoshiaki Kawasaki, Jun'ichi Kazama and Jun'ichi Tsujii. In Proceedings of SIGIR'03 Whorkshop on Text Analysis and Search for Bioinformatics. August, 2003.

• Tuning Support Vector Machines for Biomedical Named Entity Recognition.  Jun'ichi Kazama, Takaki Makino, Yoshihiro Ohta, and Jun'ichi Tsujii. In Proceedings of the Natural Language Processing in the Biomedical Domain (ACL 2002), pp. 1-8. July, 2002, Philadelphia, USA. (ps)(pdf)
• 80 abstracts used for the test in this paper are listed here.
  

• A Maximum Entropy Tagger with Unsupervised Hidden Markov Models.  Jun'ichi Kazama, Yusuke Miyao, and Jun'ichi Tsujii. In Proceedings of the Sixth Natural Language Processing Pacific Rim Symposium (NLPRS2001), pp. 333-340. November, 2001, Tokyo, Japan. (ps)(pdf)(revised ps) (revised  pdf) software

• Adaptive Morphological Analysis With a Small Tagged Corpus. Jun'ichi Kazama. Master thesis. University of Tokyo. 2001. (pdf)
  

• 「チャットのための日本語形態素解析」  風間 淳一  光石 豊  牧野 貴樹  鳥澤 健太郎  松田 晃一  辻井 潤一. 言語処理学会第五回年次大会発表論文集, pp. 509-512. March, 1999, Tokyo, Japan. (in Japanese) (ps)(pdf)