Re: [Tracker] trackerbird fixes



On 10/02/14 12:06, Michael Lipp wrote:
Hello Martyn,

Hello Michael,

I've CCd Ivan and Jürg given it involves ontology changes and they are usually quite involved in this particular area.

Am 10.02.2014 11:55, schrieb Martyn Russell:
What exactly did you want to achieve and what problems are you seeing
with the current specification?

My comment was just agreeing with Adrien in staying that technically,
the ontology is not incorrect because of the subclassing that goes on,
specifically:

   nmo:Email is a nmo:Message
   nmo:Message is a nfo:TextDocument
   nfo:TextDocument is a nfo:Document

The best way to fix this is to eliminate any subclasses in the query
in tracker-needle because technically, the spec seems right to me.

For now, I've done this in master for tracker-needle. It was simple enough:

https://git.gnome.org/browse/tracker/commit/?id=016e2dd5d9064c72b2db29340ad029d5b66b0a7b

The other approach would be to disconnect nmo:Message from
nfo:TextDocument which I don't think is the better alternative.

I think it depends on how close the tracker project wants to follow the
specification (or how far it wants to deviate from them).

We've not really been actively following the spec for a while, mainly because we had our own alterations in the earlier days and we've just not kept up to speed with them.

Which is also why I CCd Ivan/Jürg, because they've been following the spec more closely in the past.

I may be completely wrong and maybe http://www.semanticdesktop.org/
isn't the "standards body" for this stuff. But trying to understand
things better, I came across their site. And looking at

I believe it is the standards body / location indeed.

http://www.semanticdesktop.org/ontologies/2007/03/22/nmo/#Message I
found that they do NOT consider nmo:Message to be a nfo:TextDocument.
Rather, if you look at

It's entirely possible that we either fashioned the ontology to suite Nokia and their text messaging/emailing/etc requirements back in the day OR that the Nepomuk ontologies back then were less refined and things have changed. I don't know which is more accurate.

I just recall a lot of time being spent on this area of the ontology because it was vital for the N9 to have these data relationships defined well.

http://www.semanticdesktop.org/ontologies/2007/01/19/nie/#InformationElement
you find nmo:Message and nfo:TextDocument to be equal subclasses of
InformationElement. So the two are siblings, but (at least according to
that specification) nmo:Message is NOT a nfo:TextDocument.

Yes, I see.

As I said, maybe it is me not knowing enough about the Tracker project,
and the project doesn't adhere to those specifications. In that case I

We try to stick to Nepomuk as closely as possible, but we've added quite some extensions over time and it would be good to check the whole ontology really and see what should be updated.

wonder which ones are relevant for the project or whether the project
has designed its own relationships. In the later case, just to put my
two cents in, I think the specification is right. Technically, almost
everything in computing can be considered -- or at least converted to --
a text document, of course. But if I send someone a plan email, and ask
the next day "did you get the document that I mailed you?" then I
suppose at least nine out of ten people will answer "no, there was not
attachment".

I would be happy to make the change if I knew it wouldn't cause any nasty side effects.

Jürg, Ivan, any comment to add here?

--
Regards,
Martyn

Founder & Director @ Lanedo GmbH.
http://www.linkedin.com/in/martynrussell


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]