That is an unmanageable solution and it will certainly destroy the  
performance of a DHT. The technology your looking for is a binary  
search tree such as PHT, so you will need to introduce some sort of  
algorithm to map the characters of the published key to a query-able  
index. Google "DHT + PHT", or Cubit etc...

~J

On Dec 4, 2008, at 9:21 AM, Will Morton wrote:

> Hello hackers;
>
> I have a large number of items, each with string metadata, that I want
> to search for.  I want the search to be decentralised, so I'm going to
> use a DHT (although I'm happy to use another data structure if it's
> more optimal).  I want to be able to search the items based on partial
> matches, so if an item has title 'foobarbaz', I want it to be returned
> when I search for 'foo', 'bar' or 'baz'.
>
> So when I am adding entries into the DHT, do I have to add keys for
> all possible substrings of the thing I am indexing (so with minimum
> search length 3, I need to add keys for 'foo', 'oob', 'oba', etc), or
> are there optimisations I can make?  Presumably this is a well-studied
> area; can anyone point me at papers concerning this problem?  Or else,
> describe how gnutella/edonkey do it?
>
> Thanks,
>
> Will
> _______________________________________________
> p2p-hackers mailing list
> [email protected]
> http://lists.zooko.com/mailman/listinfo/p2p-hackers

_______________________________________________
p2p-hackers mailing list
[email protected]
http://lists.zooko.com/mailman/listinfo/p2p-hackers

Reply via email to