[Date Prev][Date Next][Date Index]

[ntcir:136] Seminar talks on CLIR: the 6th NII Informatics Open Forum

Dear all,

Here is a reminder for the NII Open Forum on Informatics on
"Cross-Lingual Information Access".

You are most welcome.  Please come and join the discussion!

Noriko Kando
The 6th NII Open Forum on Informatics

Date: 8th Mar, 2004     13:00-15:00
Place: National Center of Sciences 20F (Rm. 2004+2005)
  Hsin-Hsi Chen (National Taiwan University)
  Kazuaki Kishida (Surugadai University)

  -- Both are the task organizers of the NTCIR CLIR.
TALK #1:

Hsin-Hsi Chen
Professor & Chairman,
Department of Computer Science and Information Engineering
National Taiwan University

Multilingual Summarization and Translingual Transmedia Information Retrieval

In the first topic, I will introduce a general architecture of a
multi-document summarization system at first.  Several important issues
such as clustering algorithms, clustering units, features used in
clustering, and event detection will be discussed.  Then I will show how
to extend the architecture to a multilingual multi-document
summarization system.  Some extra issues like translation documents
among different languages, idiosyncrasy among languages, and user
preference will be touched on.

In the second topic, I will introduce this challenging research problem
at first.  Then, several approaches to unify the languages and media of
queries and documents will be presented.  Some preliminary results on
cross-language image retrieval via spoken query are shown.


Kazuaki Kishida
Professor, Surugadai University
NII Visiting Associate Professor

Automatic Query Refinement for Mono- and Cross-Lingual Information

One of the important techniques for cross-lingual information retrieval
(CLIR) is translation disambiguation, of which purpose is to remove
automatically erroneous translation candidates obtained in the process
of dictionary-based translation of original query terms. Kishida and
Kando (2003) have proposed a "two-stage query refinement" technique for
the disambiguation, and empirically shown that it works well at the
retrieval experiment in CLEF-2003. The author would like to talk about
procedure and characteristics of the two-stage refinement. This
technique is also easy to be applied to monolingual retrieval situation.
For example, in Japanese monolingual retrieval, it is possible that
original query is expanded by conducting repeatedly Japanese-English and
English-Japanese translations as a pseudo cross-lingual retrieval. In
order to refine the expanded set of query terms, our "two-stage
refinement" technique would be available. A result of preliminary
experiments will be presented.

This forum is open to the public.
Participation Fee : Free
Please send your name, application, address, telephone number, E-mail
address to openforum@xxxxxxxxx