NTCIR-4 CLIR
FAQ: Frequently Asked Question

[NTCIR4-CLIR]

Table of Contents

Q1. Hong Kong Standard
Q2. Document collections for NTCIR-4
Q3. CIRB011
Q4. Korean Economic Daily
Q5. Training Data

Q1. Do we use "Hong Kong Standard" File as part of the English Document Collection for the NTCIR-4 evaluation (formal run)?

A1. Yes. "Hong Kong Standard" is included in the English Document Collection for the NTCIR-4 evaluation. Please make sure you have downloaded both of the 2 Hong Kong Standard files listed below;

1. HK98+99.tar.gz: Hong Kong Standard (71.3 MB)
2. HK98+99.add.tar.gz:Hong Kong Standard (addition) (27.4 MB)

[go to top]

Q2. I want to confirm which document collections are used for the NTCIR-4 CLIR evaluation (formal run)?

A2. The followings are the document collections used for the NTCIR-4 CLIR evaluation (formal run). Please MAKE SURE that you use ALL the document collections of the language that you will test

Documents for NTCIR-4 evaluation

• CHINESE:
  • - CIRB011.tar.gz: CIRB011 (153.8 MB)
  • - CIRB020.tar.gz: CIRB020 (246.0 MB)

• KOREAN:
  • - Hankookilbo.tar.gz:Hankookilbo (93.5 MB)
  • - Chosunilbo98+99.tar.gz:Chosunilbo 98-99 (75.4 MB)

• JAPANESE:
  • - Mainichi98+99.tar.gz: Mainichi Newspaper 98-99 (127.7 MB)
  • - Yomiuri98+99.tar.gz: Yomiuri Newspaper 98-99 (225.8 MB)

• ENGLISH:
  • - EIRB010.tar.gz: EIRB010 (12.9 MB)
  • - MainichiDaily98+99.tar.gz: Mainichi Daily News 98-99 (12.9 MB)
  • - KoreaTimes.tar.gz: Korea Times (20.4 MB)
  • - HK98+99.tar.gz:Hong Kong Standard (71.3 MB)
  • - HK98+99.add.tar.gz:Hong Kong Standard (addition) (27.4 MB)
  • - Xinhua News Service file from LDC2002T31 AQUAINT Corpus

All the document collections except "Xinhua" are available from NII for the participant based on your request indicated by the user agreement forms. For the Xinhua News Service file, it will be available from the LDC after the registration through NTCIR.

[go to top]

Q3. Is "CIRB011" included for the NTCIR-4 CLIR formal run evaluation?

A3. Yes.

[go to top]

Q4. Is "Korean Economic Daily" included for the NTCIR-4 CLIR formal run evaluation?

A4. No. Since the documents in the "Korean Economic Daily"files were published in 1994, while the documents in the other document collections for NTCIR-4 are published in 1998 or 1999. As NTCIR-4 uses the comparable corpus published around the same time period in different languages, "Korean Economic Daily" was excluded from the formal run and distributed for the training purpose only. To use "Korean Economic Daily", please use the "Topics and relevance judgments for 1994" set in the NTCIR-3 CLIR collection

[go to top]

Q5. Where Can I find the topics and relevance judgments set for NTCIR-3 CLIR collection? I need it for training

A5. Formal Application is needed. Please download the forms on the following URL, complete and send it to the NTCIR Office.

http://research.nii.ac.jp/ntcir/permission/perm-en.html#ntcir-3-clir

Because of the time constraint, please fax the signed "formal application" first, then send the original copy by post.

Contact Info:

Research Center for Information Resources
National Institute of Informatics
2-1-2 Hitotsubashi,
Chiyoda-ku, Tokyo
102-8430, JAPAN

PHONE: +81-3-4212-2750
FAX: +81-3-4212-2751
Email: ntc-secretariat@nii.ac.jp

[go to top]

Last update: 2003-10-10 ntcadm-clir@nii.ac.jp