[Xapian-discuss] omega and searching specific fields

[Olly Betts]

On Sun, Mar 06, 2005 at 09:12:36PM -0500, Sig Lange wrote:
> I am indexing a music collection and want to be able to perform
> searches in certain fields.  I'm a little confused on the difference
> between "boolean filters" and "Probabilistic Fields" .. I am using
> boolean filters but perhaps use the other? I would like to use the B=
> CGI parameter to search in omega.

To put it simply:

If you want to perform a free-text search on a field (which will
generally contain multiple words of text), you want "probabilistic".

If the field contains a single category or code or similar, and you
want to be able to filter search results according to the value of
this field, you want to index it as "boolean".

> So, I came up with a index file like this:
> -- begin music.index --
> id: index boolean=Q unique=Q

You don't want index here - that will index the id as a probabilistic
term.  So unless you want someone to be able to type
"39c5236d30e6c24cb737f3d166a7e05a" into your search box, the "index"
isn't useful.

So you want:

id: boolean=Q unique=Q

> artist: lower field=artist boolean=XART index

You probably want to preserve the case of the artist in the field,
so put lower *after* field=artist.  I'm not sure if you really want
to lower for probabilistic indexing anyway...

And I suspect artist should be a probablistic field.  In which case you
want:

artist: field=artist index=XART

If you really want a boolean filter, then probably:

artist: field=artist lower boolean=XART

> path: field=url

OK.

> info: lower index

Probably just:

info: index

And I'd imagine you'd also want to index album and title (which are in
the sample record you gave).

> I can search and retrieve these documents as correctly but when I pass
> a B=XART for example, I don't seem to be getting the normal results
> back.

If XART is boolean, you need to specify a whole term, not just a prefix
- for example: 'B=XARTa perfect circle'.

If you're hoping to generate a drop-down list of artists, you'll need
to tweak Omega.  There's a commented out function in query.cc called
'do_picker' which is probably close to what's needed.  We wanted to
make it possible to generate these using omegascript, but that's not
been done yet.

Cheers,
    Olly