[Xapian-discuss] Search queries with wildcards

[James Aylett]

On Tue, Dec 14, 2004 at 03:33:27PM +0100, Timo Haberkern wrote:

> i'm looking for a search engine for a new web appliaction that we are 
> developing at the moment. I took a first look at xapian and it looks 
> very promising. The only thing i need are searches with wildcards or 
> matchin subpatterns. I test it with the index of the xapian online page.
> 
> - If i use "install" as search query i get 1 match
> - if i use "installing" as search query i get 3 matches
> 
> Is it possible that the search engine is finding all words in which the 
> pattern install is included??

I doubt it, but Olly is the best person to talk about the website
search, and probably to answer this question.

> Another question: The application is multi language. Is it possible to 
> have more than one different indexes and search a special index? And 
> that the indexing is using deiffernt stemmers ans stop words for the 
> indexing?
> 
> For example:
> 
> - A german index,  that i sindexed with a german stemmer and stop file
> - An english index,  that i sindexed with an english stemmer and stop file
> 
> If a german user is searching he will use the german database (with the 
> german stemmer for the search query)

This would be fairly straightforward, yes. Assuming the the site or
application effectively exists once for each language, you can index
into a different database for each language and then use the relevant
index, stemmer and stoplist based on which application interface
you're using - German, French or whatever. This can all be done with
Omega, except possibly custom stoplists on indexing - I think that is
still compiled in. It's not that difficult to replace the hard coding
for different languages, although a better solution would be to read
in a stop file, and would be more work.

One thing to be aware of is that because Xapian is the underlying
system, rather than the application, if you're not happy with the way
omega works you can always write your own, in C++ or in one of PHP,
Python, Perl and Java (and possibly TCL 8).

J

-- 
/--------------------------------------------------------------------------\
  James Aylett                                                  xapian.org
  james at tartarus.org                               uncertaintydivision.org