Comments on: Introducing OpenVocab

By: Oz DiGennaroן

Oz DiGennaroן — Sun, 07 Dec 2008 16:47:18 +0000

A great idea. It’s important to coordinate will all of the other efforts on the web. Well, with some of them anyway. I would like to imagine how this effort could combine with others, both technically and philosophically.
As soon as I think for a while about knowledge representation, I plunge into a morass filled with “really?”, “what does that mean?”, “is that possible”.
Exciting! What fun! And I hope and plan to make a living from this kind of stuff too.

By: Ian Davis

Ian Davis — Sat, 06 Dec 2008 20:06:39 +0000

Jonathan I’d not seen metadataregistry before. I’m investigating it in a bit more depth. Thanks for the link!

By: Jakob

Jakob — Fri, 05 Dec 2008 00:03:05 +0000

That’s a nice toy and maybe a great tool! But for the primary step of modeling there is nothing better then a whiteboard or piece of paper

Is there a similar tool for instances or can you somehow also add instances to openvocab?

By: Stuart Sierra

Stuart Sierra — Thu, 04 Dec 2008 17:57:50 +0000

Finally, the missing link! Happy to see this. I also second Taylor’s suggestion of encouraging specific types for datatype properties.

By: Taylor

Taylor — Thu, 04 Dec 2008 14:47:01 +0000

This is a good idea Ian, really good. This is the kind of activity that’s been missing with the semantic web, a more community driven shared vocabulary approach. None of this “sameAs” stuff works unless graphs eventually combine at some shared URI’s.

One thing I’d like see more of is specific ranges for properties, specifically data type properties and their XSD schema types. Much of the Dublin Core, ical, examples seem to be constructed and considered by folks who have never attempted parsing that data into a typed language. ical’s date types point to classes that have no concrete definition…you have to read the schema to know what to expect. That part of RDF/OWL really escapes me…if humans have to read documents to know how to parse a string…just doesn’t seem like what we’re trying to go for.

So it’d be interested if your tool could coax people into being a little more specific on property ranges than “resource” or “Literal”

By: Jonathan Rochkind

Jonathan Rochkind — Wed, 03 Dec 2008 15:59:52 +0000

This project seems to have a lot of overlap with Jon Phipp’s fairly mature Metadata Registry app. It’s awfully nice.

http://metadataregistry.org/

i _think_ Jon’s software is open source. I’d encourage you to coordinate/cooperate with Jon on this.

Karen Coyle and others have already done some work in Phipp’s registry on loading traditional library cataloging vocabularies and RDA-related vocabularies in.

Regarding the PDDL, I just posted something to the PDDL listserv about the need for an established standard for declaring PDDL in a machine-readable/recognizable way. There isn’t actually an acceptable identifier _for_ the PDDL, since the URL where it’s located asks you not to link to it. Hmm.

By: Ian Davis

Ian Davis — Wed, 03 Dec 2008 14:49:44 +0000

Peter,

Yes the PDDL is appropriate but it’s only a formal way to declare what I’m saying: all the data is in public domain. In reality the facts are already in the public domain but it doesn’t harm to declare it explicitly. The editing forms also have a note saying that by clicking save the user is agreeing to donate the supplied information to the PD.

There is no common practice for RDF, but that was Talis’ intention when we embarked on the work that became the Open Data Commons. We do need to be explicit about the rights asserted over data so we can safely reuse it.

By: Ian Davis

Ian Davis — Wed, 03 Dec 2008 14:46:46 +0000

Richard,

All great requests and I agree totally with them all. Some specifics:

1 shouldn’t be very hard to implement. Number 2 is also possible: all the actual triple changes are held in the platform store as RDF, I’m just not exposing them at the moment. It should be possible to do proper diffs etc.

3, 4 and 6 are valuable and I think could play into the transition between unstable/stable terms

By: Peter Murray

Peter Murray — Wed, 03 Dec 2008 14:06:04 +0000

Neat work; I'm going to play around with it a little more today. One initial observation, though, is a broken URL in the page footer for the rights link:

All text and data are in the Public Domain&;lt;/a>.

The current OCLC kerfuffle has me thinking more about licenses for data. Is there a common practice for RDF? Is Open Data Commons Public Domain Dedication and License appropriate?

By: Richard Cyganiak

Richard Cyganiak — Wed, 03 Dec 2008 11:12:08 +0000

Feature wishlist:

1. RSS feed for the recent changes to each term (I want to subscribe to the feeds for the terms I care about, so I can watch out for morons and vandals messing with them)
2. Better versioning/history; most of all I want to see what has actually changed in the history (not just the comment); and I want to be able to view older versions so I could revert to them if necessary
3. Ability to comment on terms. This is actually really important. Vocabulary development needs FEEDBACK.
4. Along with comments, I should be able to give some sort of “vote”. Maybe just a “thumbs up” or “thumbs down”. This rewards contributors and emerges high-quality, consensus-built terms.
5. Ability to attach my name to my terms, changes and comments. OpenID authentication would be a good way to do this. But mainly I want to see real names instead of just “Anonymous”.
6. Some key statistics should be big and bold on top of each term page: How long has the term existed? How many revisions? How many comments? How many up/down votes? I want to judge in a glance if a term “works” for the community.
7. Ability to cluster terms into vocabularies. This doesn’t necessarily mean a separate namespace, but rather just a way to show that some terms are supposed to (or can) be used together. A “vocabulary” could simply be a list of previously existing OpenVocab terms plus a wiki page for prose documentation and examples. If a vocabulary has been set up that contains term X, then the term page for X should have a link to the vocabulary.