About 17 years ago I work for a local council which adopted full text
indexing. All incoming correspondence was scanned and OCR as, and hand
written documents were typed and dropped into a database. Outgoing
correspondence, council and committee minutes were all added to the
database in their native WordPerfect format.

Within 12 to 18 months the problems being experienced were high recall and
low precision. People just simply gave up on using the database and went
in search of the hardcopy.

More and more today people want full-text searching and I think the ideal
solution is a combination of metadata and full text indexing.

I would like to suggest you read some of the literature from the
information science field on this subject. When I did my undergraduate
studies I read a lot of what Carol Tenopir and others had written. I have
just found there are over 1000 hits using Google with "full-text
databases" and "Carol Tenopir".

One of the things which come to mind is the usage of words. For example is
it an automobile, motor vehicle, car, vehicle, Ford, Toyota etc. I am
aware today technology can be used to search for a word which means the
same thing, for example, a 'car' is the same as 'motor vehicle'. Then
there are misspellings of words and how you address this issue in
full-text. These are the sorts of issues I read about in the information
science literature.

Cheers
Jenny Evans

List archives at http://lists.ufl.edu/archives/recmgmt-l.html
Contact [log in to unmask] for assistance