At 09:57 AM 31/03/2000 +1000, Robert A Ottaway wrote:
>I am interested in peoples views on the merits of using Imaging or OCR

>At the State Government Office (Main Roads) where I now work we use a system
>called Texpress. Documents are scanned in, OCR software is used and the
>documents are converted to text files. Searching for text is very
effective, but
>we need to continue to use all hard copy files.

I worked in a local council in the late 1980's where an imaging system was
implemented utilising OCR to allow for text searching. Inward documents
were scanned and utilising OCR the documents were turned into text, and put
into a free text database along with Council minutes, internal memoradums
and outward correspondence.

The system failed as the database grew. Precision decreased and recall was
high. Staff could not find documents. They then asked if I could help to
solve the information overload from hig recall.

What was the latest technolgy at the time failed. The efficency gain with
records staff losing their jobs did not eventuate. The paper won out in the
long term because consideration was not given to retrieval aspects or the
ability to view a sequence of transactions.

My view is by all means utilise free text at the lower end through
documents. However at the broad end you need to have an electronic file
titled using a classification system. The documents should be attached to
the electronic file as each transaction occurs. You also need to capture
metadata about the document to make it searchable as it does not take very
long to find out people who write into an organisation can have problems
with spelling, and you will also find they use different terminology to
describe something, eg car, motor vehicle, Ford, Falcon...

The other thing you need to consider with OCR and image is the legalities.
If you are not retaining the exact original as an image then you would
require the original document for evidential purposes. If you retain as
image then you need to document the process or the evidence may not be

Jenny Evans