On Sun, Aug 20, 2000 at 02:48:43PM +0300, Itamar Shtull-Trauring wrote:
> http://paris.cs.berkeley.edu/~dawnsong/papers/se.ps
> 
> Useful for searchable meta-data?

It is very cool paper, the gist of the idea being that you xor every word
in the text with a number from a crypto stream followed by a number
generated from (by say a hash) the preceding number and and the plaintext
word encrypted by a key you keep secret. Then you can send the ciphertext
to somebody and ask them to check if a word is in the text by sending the
word and the word encrypted with that secret key - so that they can xor
every word in the cyphertext with the search word and see if it produces a
pattern of <number>+hash(<number>+<encrypted word>) in which case it is a
hit. (If you don't want the person to know what you are searching for you
start by encrypting the data word by word with a blockcipher in a mode
that produces each block independent of the others).

It isn't of much help for us though, since somebody searching for data is
not going to have the secret key to encrypt the search word. And it only
matches on whole words, so, since we don't have to worry about being able
to restore the entire data on searches, we could also just hash every
search term.

> 
> -- 
> Itamar S.T.  itamar at maxnm.com
> Fingerprint = D365 7BE8 B81E 2B18 6534  025E D0E7 92DB E441 411C
> 
> _______________________________________________
> Freenet-dev mailing list
> Freenet-dev at lists.sourceforge.net
> http://lists.sourceforge.net/mailman/listinfo/freenet-dev
> 

-- 
\oskar

_______________________________________________
Freenet-dev mailing list
Freenet-dev at lists.sourceforge.net
http://lists.sourceforge.net/mailman/listinfo/freenet-dev

Reply via email to