Re: [Tracker] Tags, metadata and date fields

[Adrien Bustany <abustany gnome org>]

On Sun, 04 Apr 2010 20:20:06 -0400, Nikolaus Rath <Nikolaus rath org>
wrote:
> On 04/03/2010 09:56 AM, Adrien Bustany wrote:
> > > I am currently using a set of self-developed scripts to manage a
> > > document archival system. When a document is added to the archive,
> > > the user has to assign a name, tags from a predefined list, a date and
> > > optionally a couple of keywords. The script then extracts the plain
> > > text from the document, adds the additional metadata and writes
> > > out an XML file. The XML file is then indexed by Swish++.
> > >
> > > For retrieval, another script asks for keywords, tags and a
date-range,
> > > invokes Swish++ to do the search and retrieve the metadata, presents a
> > > nice list with results and finally opens the original document.
> > >
> > > I was wondering to what extent it would be possible to replace this
> > > system with something more established like tracker. From the
> > > documentation, I gather that it is possible to assign tags to the
> > > documents. But can I also add additional meta data, like title,
keywords
> > > and timestamps and show them in the result list? Is it possible to
> > > constrain the results to a given time span?
> >
> > Yes, Tracker provides ontologies (= metadata standards) for tags, date,
> > keywords and stuff. Tracker uses the Nepomuk ontologies
> > (www.semanticdesktop.org/ontologies/) to describe its data.
> > The queries are then done with a langage called SPARQL. SPARQL is a
very
> > powerful languages, and allows you to of course provide dates
> > constraints,
> > but also any type of constraints. I invite you to look at a SPARQL
> > introduction, or at the Tracker documentation for that
> > (http://live.gnome.org/Tracker/Documentation).
>
> The querying with SPARQL looks quite promising indeed, thanks for the
> quick answer!
>
> However, I wasn't really able to find any information on how to get the
> metadata into my index in the first place. Could you give me some
> information where to start? Do I move the documents into a
> tracker-watched folder and manually modify the database after the
> documents have been indexed? Or do I explicitly ask tracker to add a
> specific file with specific metadata to the index? Or do I add the
> metadata as extended attributes of the file and put it in a watched
> folder? Or something else entirely?
>
>
> Best,
>
>    -Nikolaus

Information is entered in Tracker either by "miners" or directly by
applications.
The miners are small modules that crawl data sources (for example, your
filesystem)
and insert data in Tracker. Currently Tracker ships with two miners, one
crawling
the FS and another the installed applications. So basically what you have
to do is
configure the FS miner to ensure it's crawling the right dirs for you.
If you want to add custom triples in the store, you can use the
tracker-sparql tool.

Cheers

Adrien