2003 Fiscal Year Final Research Report Summary
Automatic Transformation of GDA Document Tag and Development of Its Applications
Grant-in-Aid for Scientific Research (B)
|Allocation Type||Single-year Grants |
|Research Institution||KYOTO UNIVERSITY |
OKUNO Hiroshi Kyoto University, Graduate School of Informatics, Professor -> 京都大学, 情報学研究科, 教授 (60318201)
HASHIDA Koiti Advanced Institute of Science & Technology, Cyber Assist Res., Ctr, Vice-Director, サイバーアシスト研究センター, 副所長
SATO Satoshi Kyoto University, Graduate. School of Informatics, Associate Professor, 情報学研究科, 助教授 (30205918)
KAWAHARA Tatsuya Kyoto University, Academic Center for Computing and Media Studies, Professor, 学術情報メディアセンター, 教授 (00234104)
KAMATANI Kazunori Kyoto University, Graduate School of Informatics, Research Associate, 情報学研究科, 助手 (40362579)
|Project Period (FY)
2001 – 2003
|Keywords||Document Tags / Global Data.Annotation (GDA) / MPEG-7 / Semantic Description Scheme / Lecture Minute Indexing / Linguistic Description Scheme / MPEG-7 Music Descriptor / Privacy-enhanced Access Control|
We have obtained the following results concerning GDA (Global Data Annotation) standardization, applications and peripheral technologies.
(1) GDA annotation of Mainichi Newspaper articles: 200 articles of Mainichi Newspaper of the past 10 years are manually annotated with GDA tags including anaphora and co-references.
(2) Proposal of GDA concepts to MPEG-7 Standardization: The proposal or Linguistic Description Scheme based on GDA and UNL (Universal Network Language) has been adopted as MPEG-7 Second version Working Draft.. In addition, it is under negotiation by ISO/TC37/SC4 Language Resource Management Standardization. A simple automatic transformation from GDA to UNL enables to translate a simple GDA-annotated. Japanese sentence into English, Italian, Spanish and Arabian automatically.
(3) MPEG-7 annotation for music: Vie pointed out the problems of.ontology alignment that is well-known to AI community, and generated automatically a musical instrument ontology by which known and unknown musical instruments are identified with a high recognition performance.,
(4) Automatic speaker indexing of minutes of round table: For round tables with alternative speakers, automatic speaker indexing has been developed with multi-speaker models obtained by training with a very large dialogue corpus, domain knowledge obtained by a very large minutes corpus of the Lower House. Thus, its performance is quite high.
(5) Privacy-enhanced access control: Although security draws much attention, privacy is never-the-less important for accessing various corpora. For this purpose, we designed a privacy-enhanced access control based on the SPKI (Simple Public Key Infrastructure) and demonstrated its feasibility by implementing a Web server.
Research Products (14 results)