Xapian discuss - Oct 2005 - How to deal with "zones" on documents

Hello,

I'm a newbie on Xapian, and after installing it (easyly, so thanks to
development team) I have some questions.

I want to have a different weight on result for a term, if it comes from
differents parts of a document.

For example, on HTML files, from tags <title>, <h1>, <h2>, ...

Must I do it with different database (one per "zone") and make a query
with all databases ?

Is there a doc on that, or a thread on mailing list archive, to read ?

I test Xapian to use it on a search engine for the collaborative software I work
on (Mioga2 : http://www.mioga2.org).
For this time, we use mifluz from Senga, but it is a dead project.

Best regards

Gilles Polart-Donat

PS : I'm a french guy with poor english, sorry.  ;-)

On Sat, Oct 29, 2005 at 12:40:40PM +0200, Gilles Polart-Donat
wrote:
> I want to have a different weight on result for a term, if it comes from 
> differents parts of a document.
> 
> For example, on HTML files, from tags <title>, <h1>,
<h2>, ...
Assuming you know the relative weights you want to apply to the
different tags, then just apply extra wdf to terms generated from these
fields.  You do this by passing a value greater than 1 for the optional
third argument to Document::add_posting() (or the optional second
argument to Docuemnt::add_term()).

E.g.

    doc.add_posting(term, pos, wdfinc);

    doc.add_term(term, wdfinc);

So "wdfinc" is the "extra weight factor".  Note that it must
be an
integer.

If you want to be able to tune the factors dynamically, you could
index terms from particular tags with particular prefixes (e.g.
the term "xapian" in <h1> might be XH1:xapian).  Then for each
term in the query, you'd produced an OR of the various possible
forms with the wqf (within query frequency) set appropriately
for each form:

(xapian OR XH1:xapian OR XH2:xapian OR ...)

The static way is likely to be more efficient, so use that if you
can.

Cheers,
    Olly