On Sunday 26 September 2004 01:15, Erik Moeller wrote:
If you're interested in the Wikidata project but
you can't directly
contribute in terms of code, you can help with some research:
http://kendra.org.uk/
http://dev1.kendra.org.uk/
These people are doing something very similar, i.e. building free-form
databases using a wiki-like model (see the second link).
We need to know:
1) What exactly are Kendra's capabilities?
2) Is there potential for cooperation between Kendra and Wikimedia?
If you want to research this, please respond to this message so that
there's no needless duplication of effort. You could use
http://meta.wikimedia.org/wiki/Kendra_evaluation
to chronicle your findings.
I couldn't find anything on that page, so I'm posting here. I hereby release
this e-mail under GFDL so anyone can use it to start a page on Meta or
anywhere else.
Capabilities... uhh. It lets you freely edit pages. Pages may contain ordinary
text, links, in the form of [[link]], what I'll call equivalences, in the
form of [[key = value]], and what I'll call definitions, in the form of
[[key; value]].
The equivalences let you set property of a page. Currently defined properties
are:
is_a
has_property
has_range
has_option
See
http://dev1.kendra.org.uk/kendrabase/semantic_property for more info on
this. Most example pages have an "is a" property while some have and some
don't have more properties. AFAICT, you can think of properties as about
table column definitions.
Their examples start with an ontology:
music_ontology:
----
[[is a = ontology]]
This is the bare bones of an ontology for [[music]]:
* [[album]]
* [[song performance]]
* [[artist]]
* [[track]]
----
What's an ontology for I can't tell. You can make something similar to a table
definition:
album:
----
[[is a = musical work]]
* [[has property = album title]] -- ''not sure what this means''
* [[has property = artist]]
* [[has property = year]]
* [[has property = publisher]]
* [[has property = song performance]]
----
Each of these properties is defined on own page. I like this idea, though I
see a lot of potential for naming conflicts here. Few examples:
artist:
----
[[is a = human being]]
----
year:
----
[[has range = integers]]
*[[has option = 2000]]
*[[has option = 2001]]
*[[has option = 2002]]
*[[has option = 2003]]
*[[has option = 2004]]
*[[has option = 2005]]
*[[has option = 2006]]
----
Kendra will let you easily define a query, for example see
http://dev1.kendra.org.uk/cgi-bin/bootstrap.sh?action=query&title=albumlbum. I
don't see a way to easily link to a query or include it into an article.
Finally, an example of actual data:
yes again:
----
[[is a; album]]
*[[album_title; yes_again]]
*[[artist; yes]]
*[[publisher; huge_records_co]]
*[[year; 2000]]
----
From what I can tell, it seems that it doesn't have wiki features (page
history, watchlist, current events), but this of course shouldn't be too hard
to implement and I suppose that it's in plan.
They are using UseMod for their wiki, and Kendra is made in Python. So I don't
think that there is much possibility for direct cooperation - surely, ideas
could be exchanged, but Kendra can't just be poured into MediaWiki. OTOH, the
Kendra site is done in PHP, so probably some of their developers know PHP.