.P/taxhol.txt 

Previous experience tied to the explotation/marketing of SR&ED results.

Gary Perlman's research and experiemental development
is focused on ways to better exploit the assets of the OCLC Library Cooperative,
specifically reference and resource sharing services
such as the FirstSearch research service
centered on the WorldCat database.
WorldCat contains 50 million bibliographic
records of materials in libraries worldwide
and is the cornerstone of OCLC's current strategic plan.
Enhancing the usefulness of the information in WorldCat
will increase the markets OCLC can serve,
and increasing the multilingual features will enhance
the market of OCLC product and services internationally.

Working at OCLC's main site for the past seven years,
Gary Perlman has pushed FirstSearch to the leading edge of
library research tools.  In Quebec, he will continue this work,
and also be able to leverage the multilingual strengths of
the area to further advance OCLC's global markets.

Gary Perlman, through his work on the HCI (Human-Computer Interaction) Bibliography
(www.hcibib.org) and more recently on OCLC FirstSearch (firstsearch.oclc.org),
is a leading expert on the development of bibliographic and full text research tools.
Through his work on the HCI Bibliography and in the area of HCI,
Gary Perlman is one of the world's most widely recognized researchers in HCI,
and is able to use his central role to draw on resources worldwide.
As a researcher and software developer, he is able to incorporate leading-edge
results in systems that people can use in the very near future;
technology transfer to Web-based search engines happens in months
from conception, to analysis, through internal and external evaluation.

Specific examples:
Applying his knowledge of
computer science
(e.g., using hash tables of hash tables for fast lookup translations of structured terminology),
and French (for preliminary translations, but primarily of the structure of European languages),
the candidate initiated the creation of the first multilingual
version of FirstSearch, built the production system and performed the performance analysis.

Applying knowledge of expert systems and online help systems,
the candidate developed an expert system to identify problem areas in search
(e.g., too many or too few matches),
diagnose the reasons for the problems
(e.g., too few terms or possible spelling error),
and offer suggestions to improve results.
In many cases, problems were identified by the analysis of search logs
for millions of searches, and in all cases,
the effectiveness of the the search results expert system
was evaluated through the analysis of millions of subsequent searches.

holdings, doctpye tabs and material type - previously unexploited

multilingual FirstSearch 4.0
search history
FirstSearch 5.0 architecture
Dr. Search and spelling
Library ranking - deciding when to do
Unicode display
Document type tabs and subtypes - what to show and when to do
More like this, title+author, FRBR
FS redesign of basic search
hot topics
OPAC links
suggest-o-matic

ILL integration
Group Catalog
FRBR

HCIBIB
|STAT
questionnaires


Measures: 

number of searches, percentage of searches
normalized by logins
reduced response time
reduced zero hit search
increased holdings
amount of data used
# limit links
# uses
change in usage