[ntcir:243] Seminar: July 20, Vocabulary, Statistics, Time and Geography

["kando@xxxxxxxxx" <kando@xxxxxxxxx>]

Dear All,

We are pleased to invite you to a seminar on *Vocabulary, Statistics,
Time and Geography*. Dr Frederic C. Gey (University of California,
Berkeley / Visiting Researcher, NII) will present his recent research.

You are most welcome to join us!


**********************************************************
                     Seminar Talk on
Vocabulary, Statistics, Time and Geography

by Dr. Fredric Gey
(University of California, Berkeley / Visiting Researcher, NII)

Schedule: Friday, July 20th, 2007 (14:00-15:30)
Location: National Institute of Informatics (Tokyo),
          1st floor, Special Conference Room (Room 103)
          $BFCJL2q5D<<(B 103
Language: English
Registration fees: None


Abstract:
Official government statistics are usually identified by time and
geography and hence amenable to display by GIS.  The connection of
textual documents to time and space is less obvious.  However, news
stories usually are about a specific event in a specific place (such as
an earthquake in Iran or a shark attack off California).  In addition,
books cataloged in university library catalogs are sometimes identified
by place which is the subject of the book (i.e. a history of Tulare
County California).   The presenter's research began as a vocabulary
module to improve search effectiveness and evolved into a multi-genre,
unified search portal for statistical data coupled with finding books
about the geographic place being searched for.  We have constructed
demonstration systems accessing census data for California as well as
historical census data for the United States, together with book and
Wikipedia search.

The second part of the talk will concern itself with the problems of
evaluating geographic search upon text, specifically news stories.   How
does geographic search differ from ordinary search?  I.e. how to
evaluate search for queries like:
$B!&(BFind stories about cities along the Rhine and Danube rivers.
$B!&(BFind stories about places within 100 kilometers of Frankfurt Germany
The speaker has organized evaluation of geographic text search in
European languages (English, German, Portuguese and Spanish) as part of
a European digital library initiative to objectively evaluate search and
will discuss the role of gazetteers in geographic search.


About the speaker:
  Fredric Gey is an information scientist at the University of
California, Berkeley, where he does research into multilingual
information access, social science information systems and multi-genre
search.  His PhD disseration was on probabilistic ranking algorithms for
document search.  He has received numerous research grants from the US
National Science Foundation, DARPA, and Institute for Museum and Library
Services.  He has participated in all six NTCIR workshops, most recently
concentrating on Chinese-English cross-language search.  He was General
Chair of the 1999 SIGIR Annual Conference on Research and Development in
Information Retrieval.