Path: utzoo!utstat!jarvis.csri.toronto.edu!mailrus!iuvax!watmath!looking!brad
From: brad@looking.on.ca (Brad Templeton)
Newsgroups: news.software.b
Subject: Re: Not Checksums, how about string search hash codes?
Message-ID: <50572@looking.on.ca>
Date: 21 Nov 89 04:11:26 GMT
References: <49602@looking.on.ca> <1989Nov20.073448.18953@twwells.com>
Organization: Looking Glass Software Ltd.
Lines: 15
Class: discussion

In article <1989Nov20.073448.18953@twwells.com> bill@twwells.com (T. William Wells) writes:
>This is easy enough. Just take any hashing function and use it to
>set bits in a bit table for each word in the article. If a hashed
>word creates a bit that is not in the table, the word is not in
>the article.

Is that all it is?  Seems that would require a lot of bits.  There were
151 unique words in your article, so you would need a 1500 bit table
(188 bytes, or 226 bytes of printable characters) to get a 10%
hit ratio using this algorithm.

Of course you could have the hit ratio vary according to the article,
but 226 seems like a lot.  I thought there were more compact algorithms.
-- 
Brad Templeton, ClariNet Communications Corp. -- Waterloo, Ontario 519/884-7473