Wendy Sudbury writes:

>    Could colleagues share their experience of successful use of OCR
>    software?  We have A5 index cards, typed in assorted fonts,  4 or 5
>    columns across, with a few handwritten markings (numbering and
>    ticks in pencil or biro).  I'm looking to migrate the information
>    to a table in Word whenever we have to update a card.  We update
>    modest quantities, maybe 40 a month.  It seems to make sense to
>    scan the old card first.

Wendy

Like most other respondents, I am rather pessimistic.

I have used OCR with moderate success over the last 10 years or so but
it has only really been cost effective when working from well structured
lists done on an electronic typewriter or printed catalogues.  I did
successfully complete one project which involved scanning manually typed
accession register pages with almost 10,000 records but I only scanned
the descriptions.  Trying to get the software to tabulate all the other
columns correctly was a nightmare as the spacing was variable so I just
masked and scanned the descriptions and then re-keyed the rest of the
data.

I encountered all the problems with l/1, m/ni, S/5/s, etc. that Chris
Meaney has described and by the time I had masked the description column
for each page, scanned it and resolved the misread characters the value
of OCR was very marginal.  Indeed, now that I have found a good data
entry typist who understands museum data, I would not go down the OCR
road again for typescript of that quality.

From the sound of your data (small cards, typed in assorted fonts, 4 or
5 columns across, handwritten annotations which will probably come out
as random 'noise') I would suggest that OCR will take longer than
retyping.  So, I'm afraid I must agree with everybody else that OCR is
unlikely to be a viable solution.

Incidentally, your thoughts might now turn to voice recognition.  I have
had some moderate success with this but on the whole I have not found it
well suited to collections management data.  It works best with flowing
prose so it might just be viable for descriptions but for analysed data
it tends to be unreliable in my experience.  Admittedly, I don't have
the very latest software but I suspect that good data entry typists will
still be able to find work for a few years more.

Sorry to be so pessimistic.

Best wishes

Stuart
------
Stuart Holm, Heritage Documentation Projects            Tel: 01603 870772
2 New Road, Reepham, Norwich NR10 4LP                   Fax: 0870 055 3623
                      E-mail: [log in to unmask]
-------------   World Wide Web - http://www.holm.demon.co.uk   -------------