Recent Developments of COCOSDA-A Progress Report

 

Lin-shan Lee

National Taiwan University, Taipei, Taiwan

Republic of China

lsl@iis.sinica.edu.tw

 

 


Abstract

COCOSDA is an international organization for coordinating the globalized efforts in language resources and speech technology evaluation. Currently, COCOSDA is organized with a structure which reflects the two dimensions of its functionalities: "Topic Domains" and "Regional Programs". The former considers the dynamic technology environments, while the latter addresses the regional differences and activities. Four topic domains have been established: Evaluation of Speech Understanding/Dialogue Systems, Multi-Modal Corpora, Corpus Annotation Tools and Local Languages. Six regional programs are currently present: North America, Europe, Asia, Oceania, Latin America and Africa. In this report the recent developments of COCOSDA, starting with the previous COCOSDA Workshop at Eurospeech, Budapest 1999, including all relevant activities and programs and the current organization of the Central Coordinating Committee (CCC), will be briefly reported.

 

I. Introduction

 

The goal of COCOSDA is to set up, encourage and support globalized coordination and cooperation in developing spoken language resources and technology assessment methodologies. Collaboration in these areas which transcends national boundaries is important both because of the scientific value attached to such systematic work encompassing large number of languages and analytical approaches, and also because of the practical need to establish common methodologies for technology performance description and quantitative comparison The initiative for such international efforts came about as the result of a series of meetings:

- March 1982, Gaithersburg, USA

- September 1989, Noordwijkerhout, the Netherlands

- November 1990, Kobe, Japan

- September 1991, Chiavari, Italy

Starting 1992, formal COCOSDA Workshops have been held yearly as satellite events as ICSLP in even years and Eurospeech in odd years. These include:

October 1992, Banff, Canada, ICSLP'92

September 1993, Berlin, Germany, EUROSPEECH '93

September 1994, Yokohama, Japan, ICSLP'94

September 1995, Madrid, Spain, EUROSPEECH '95

October 1996, Philadelphia, USA, ICSLP'96

September 1997, Rhodes, Greece, EUROSPEECH '97

November 1998, Sydney, Australia, ICSLP'98

September 1999, Budapest, Hungary, Eurospeech '99

These annual workshops have been serving as forums for informal presentation of the state of the art, discussions, and decisions on actions to be taken during the following year.

 

II. Recent Developments

 

During the Budapest meeting held in Sept 1999 at Eurospeech, it was proposed that COCOSDA should organizetopic domainsand regional Programsas the two dimensions of its functionalities. Thetopic domainsreflect the global issues including the dynamic technology environments, whileregional programsaddress the regional differences and activities. Eachtopic domainorregional programcan be organized by a rapporteur, whose responsibilities include organizing presentations or sessions in COCOSDA Workshops. Some possible titles for topic domainswere suggested. It was also suggested that the Central Coordinating CommitteeCCCof COCOSDA should be re-organized, and some possible names for convenor , deputy convenor for 2000-2001 were suggested.

In March 2000, a few key officers were decided via e-mails. Lin-shan Lee at National Taiwan University , Taiwan was assigned the duty as the convenor, and Khalid Choukri of ELRA , France as the deputy convenor. Two new topic domains were then decided via e-mail discussions: the topic domain ofEvaluation of Speech Understanding / Dialogue Systemshas Wolfgang Minker of Daimlerchrysler , Germany as its rapporteur, while the topic domain ofMulti-modal Corporahas Satoshi Nakamura of ATR , Japan as its rapporteur. Three regional programs were also initiated, Khalid Choukri of ELRA, France as the rapporteur of Europe, Shuichi Itahashi of University of Tsukuba , Japan as the rapporteur of Asia, and Bruce Millar of Australian National University , Australia as the rapporteur of Oceania. Great efforts were also made in organizing sessions in ICSLP 2000 on Language Resources and Technology Evaluation, including a special session plus a regular session.

Much more actions were then taken during the Athens meetings held May/June 2000 at LREC conference. The Major conclusions of the meetings are summarized below.Corpus Annotation ToolsandLocal Languagescould be two new topic domains, and possible rapporteurs were suggested. The term of  "local languages" is to replace the previously used term ofminority languages. Although many local languages are spoken in different regions , many issues regarding local languages are common across different regionse.g. experiences in collecting corpora , technologies for developing applications , and social , cultural , economic considerations, which could be addressed by this topic domain. Two new regional programs for Africa and Latin America could be established, since significant activities in collecting corpora and developing applications in these regions have been observed, and possible rapporteurs were suggested. Moreover, further approaches were discussed for making efficient use of the Web for COCOSDA activities to be accessed by the global research community.

Much more actions were then taken during June - Sept 2000.Two new topic domains were actually developed via e-mail discussions. The topic domain ofCorpus Annotation Toolshas Steven Bird of LDC , USA as its rapporteur, and the topic domain of Local Languageshas Dafydd Gibbon of University of Bielefeld , Germany as its rapporteur. Two new regional programs were established via e-mail discussions as well. Justus Roux of University of Stellenbosch, South Africa as the rapporteur for Africa, and Elsa Mora of University of Los Andes, Venezuela as the rapporteur for Latin America. The session organizationincluding a special and a regular sessionsin ICSLP 2000 was finalized. Generic Questionnaires were developed for topic domains and regions to collect information from the research community worldwide. Great efforts were also made trying to improve the COCOSDA Website. The COCOSDA Workshop and relevant activities in ICSLP 2000 were organized.

 

V、Current Central Coordinating CommitteeCCC

 

The Central Coordinating CommitteeCCChas been re-organized. The Committee has two parts, a working group and an advisory committee. The CCC members as of Oct 2000 are listed below:

(T)Working Group

·Officers

convenorLin-shan Lee , Taiwan

deputy convenorKhalid Choukri , France

secretaryNick Campbell , Japan

· Topic Domain Rapporteurs

Evaluation of Speech Understanding / Dialogue SystemsWolfgang Minker ,Germany

Multi-modal CorporaSatoshi Nakamura , Japan

Corpus Annotation ToolsSteven Bird , USA

Local LanguagesDafydd Gibbon , Germany

· Regional Program Rapporteurs

EuropeKhalid Choukri , France

AsiaShuichi Itahashi , Japan

OceaniaBruce Millar , Australia

AfricaJustus Roux , South Africa

Latin AmericaElsa Mora , Venezuela

· Data Center Representatives

LDCSteven Bird , USA

ELRAKhalid Choukri , France

(U)Advisory Committee

· Adrian Fourcin , UK

· Hiroya Fujisaki , Japan

· Akira Kurematsu , Japan

· Mark Liberman , USA

· Joseph Mariani , France

· Louis Pols , Netherlands

 

W、More Information

 

More information about COCOSDA is available at the Website below:

http://www.itl.atr.co.jp/cocosda/