[Xapian-discuss] Indexing & Querying multiple fields

[Olly Betts]

On 17/01/07, M.J. van der Veen <forward at mvanderveen.nl> wrote:
> However, given these indexed words for 1 record:
> Term List for record #1: D19830218 M198302 Q2 Rbeste Rbritanny Rde Rik
> Rlievelingssport Rmijn Rpiano Rroeien Rtophit XC:Rgroningen
> XCgroningen XGm XNmaarten XP1 XPR1 XS20 XUmaarten Y1983 band beste
> britanny de dit idool ik is lievelingssport mijn phantom piano profiel
> roeien sjonnies speel titlesong tophit van zelf
>
>
> How is it possible that I can succesfully find this record by searching with:
> P=pianno
> DB=profiles
> FMT=searchProfiles.xml

I assume you mean "P=piano"?

> And the search fails with:
> P=britanny
> DB=profiles
> FMT=searchProfiles.xml

My guess would be that you're stemming the query with a different
stemmer to that used for the query, and "piano" stems the same way but
britanny doesn't.  You can add $querydescription to the FMT template
to show the query which actually is being run.  The index time
stemming is set by passing "-s LANGUAGE" to omindex or scriptindex,
while search time stemming is set by $set{stemmer,LANGUAGE}" (see
docs/overview.txt for more information).

> Is the delve tool actually accurate about the indices? Is there
> another way to view them?

Yes, delve is accurate.  You can also use FMT=godmode with omega to
view much the same information in a more interactive way.

Cheers,
    Olly