Message-Id: <mailto:199410310028.SAA20314@library.wustl.edu> Date: Mon, 31 Oct 1994 00:24:23 +0000 From: Frank Dunn <mailto:frank@BRAZEN.DEMON.CO.UK> Subject: Re: EXCALIBUR ?? To: Multiple recipients of list IMAGELIB <mailto:IMAGELIB@ARIZVM1.BITNET>
snip>...We currently have an imaging pilot project using Excalibur
>and I would be interested to hear from anyone who has used this product
>either successfully or unsuccessfully. I am not crazy about its text
>retrieval capabilities and feel that fuzzy searching is just not enough
>(though it's useful when combined with OCR). Any comments you may have
>would be much appreciated.
>Thanks
>Julia Urwin
mailto:>mailto:A5_JURWIN@NELINET.ORGIt was in our initial cut for a text OCR and imaging project. The main problem we saw was that as OCR isn't near 100% accurate - 98% isn't good enough - and our text data is sold on, then any host would also have to have the fuzzy logic technology. That made it a bit of a non starter, as the data with an expected percentage of noise in it becomes non-transferable, unlike clean ASCII.
mailto:Frank@brazen.demon.co.uk News International Newspapers Ltd., mailto:Frank_Dunn@delphi.com 1 Virginia Street,London,UK, E1 9BD. CompuServe:100012,23 +44 171 782 6384/6 +44 171 782 5213 fax "It must be jelly 'cos jam don't shake like that"