RE: Critical Decisions

Date view Thread view Subject view Author view

From: Dan Kohn (dan@teledesic.com)
Date: Sun Mar 12 2000 - 22:01:20 PST


I'm a business in Seattle and would like to get 400 pages entered with 99.9%
accuracy within 48 hours. Any transfer of paper to Uganda makes things
basically infeasible, because you have a 5-day ONE WAY FedEx trip, and the
cost probably overwhelms Uganda's low labor costs.

The solution has to be to FedEx the documents to a document scanning house
in the States. WorldDataNow.com can eventually create a US office for the
purpose, but in the meantime, we should probably just outsource.

The documents are scanned and converted to gzipped TIFFs (or maybe PNGs?),
which are HTTP'ed to Uganda. The originals are FedExed back or shredded.
In Uganda, one team applies human correction to OCR output, while a second
team hand keys from scratch. The third team rectifies the two outputs into
the final document which is emailed to the company in Seattle.

Don't you think human assisted OCR diffed with hand-keying will be more
reliable than either approach alone (or either approach done twice)?

                - dan

--
Daniel Kohn <mailto:dan@dankohn.com>
tel:+1-425-602-6222  fax:+1-425-602-6223
http://www.dankohn.com 

-----Original Message----- From: Gavin Thomas Nicol [mailto:gtn@ebt.com] Sent: Sunday, 2000-03-12 19:03 To: 'Dan Kohn'; 'Steve Nordquist'; 'FoRK' Subject: RE: Critical Decisions

> Who could I outsource the scanning to in the US and Europe if > I wanted to be able to overnight documents there, the TIFFs forwarded by > email, and then have all of the OCR and data entry done in Uganda?

You'd be better off xeroxing the pages, and then having someone in Uganda/wherever double-key everything. With reasonably sophisticated DTD's, I've seen 95% tagging accuracy from folk in the Phillipines.

Even top of the line OCR isn't really cost effective yet.


Date view Thread view Subject view Author view

This archive was generated by hypermail 2b29 : Sun Mar 12 2000 - 22:14:15 PST