Ticket Navigation

Ticket #80 (new defect)

Opened 2 years ago

Last modified 1 year ago

ocr fuzzy-matching misclassifying "committment" as "million"

Reported by:	jason.haar@trimble.co.nz	Assigned to:	decoder
Priority:	minor	Milestone:
Component:	Image Analysis	Version:	3.5.1
Keywords:		Cc:

Description

We got a FP today where a mail containing a JPG of a scanned page got pinged as spam by FuzzyOCR.

There were no drug/stocks related words in it, but fuzzyocr found "million", "thousands" and "alert" in it anyway. Must have been via fuzzy-text matching?

I realise this is one of the risks, but maybe there's some way of improving that?

BTW "gocr -l 180 -d 2 -i " pulled almost the entire paragraph from the page word-perfect (pretty good tool that :-)

Thanks

Jason

Attachments

Change History

08.07.2007 10:21:49 changed by anonymous

wine cellar wine gift

11.07.2007 14:32:04 changed by anonymous

http://corp.cwsurf.de/game/adult-free-game/adult-free-game-online-xxx.html adult free game online xxx http://corp.cwsurf.de/game/adult-free-game/adult-free-game-online-porn.html adult free game online porn

11.07.2007 19:27:03 changed by anonymous

http://www12.asphost4free.com/allmask/game/disney-game/2006-channel-disney-game-hot-so-summer.html 2006 channel disney game hot so summer http://www12.asphost4free.com/allmask/game/disney-game/disney-channel-lizzie-mcguire-game.html disney channel lizzie mcguire game http://www12.asphost4free.com/allmask/game/disney-game/channel-cody-disney-game-zack.html channel cody disney game zack

17.08.2007 18:42:30 changed by anonymous

Baseball Training Drill Free Fantasy Baseball Mock Draft

Add/Change #80 (ocr fuzzy-matching misclassifying "committment" as "million")

Your email or username:

Comment (you may use WikiFormatting here):

Download in other formats:

Comma-delimited Text
Tab-delimited Text
RSS Feed