Re: Boolean Query Algorithm

Sherrie Laraurens wrote:
> Hi all,
>
> I have a question relating to how search engines and (in fact
> anything else that supports boolean queries) manage to do such
> things so efficiently.
>
> my question involves a query were you would like to retrieve the all
> the documents in your corpus that have the words "Cat" and "Bat" in
> them and that they not only contain both words but that they must be
> consecutive for example "Bat Cat"
>
> I can think of a very crude way of doing this which involves hashing
> every word in a document into a hash table and storing an index for
> said document , then in the query stage to hash both words (hash
> join) get the intersection vector of the resulting vectors from the
> hashing process, then one by one examine each document from the
> intersection vector find the word "Bat" and see if the word "Cat" is
> the next word if it is place said document into the final result
> vector and once finished pass back to user.
You may want to check out the following technology...It is far from being perfect but it is well targetted at better support of text searches...It may give you some ideas even though several things concerning RM are false...

http://www.armadillo.fr/english/whitepapers/WHITEPAPER_2004.htm Received on Wed Jul 12 2006 - 14:00:20 CEST