Mapping between SIOC and EvoOnt, some other comments on EvoOnt

23 views
Skip to first unread message

Matthias Samwald

unread,
Oct 28, 2008, 6:47:41 AM10/28/08
to baetle
I am looking at the latest version of the EvoOnt ontologies that can
be found at [1]. I think there is potential for a more extensive
mapping to SIOC.

bom:Comment could be a subclass of sioc:Post
bom:Issue seems to me like a subclass of sioc:Post, rather than
sioc:Container.
bom:User is quite redundant with sioc:User. I think that rather than
creating a new subclass, one could just add additional properties to
sioc:User.

Generally, I also suggest to add rdfs:labels (with a 'en' language
tag) to all entities in EvoOnt, as many user interfaces depend on the
existance of these labels. The ID part of URIs should not be used for
human-readable labeling.

It also seems like the bug in the bug ontology still exists: bom:Issue
is currently a subclass of two disjoint classes, which makes the
ontology inconsistent.

[1] http://www.ifi.uzh.ch/ddis/evo

Cheers,
Matthias Samwald
Semantic Web Company, Vienna

Jonas Tappolet

unread,
Oct 28, 2008, 12:29:04 PM10/28/08
to bae...@googlegroups.com


> -----Original Message-----
> From: bae...@googlegroups.com [mailto:bae...@googlegroups.com] On
> Behalf Of Matthias Samwald
> Sent: Tuesday, October 28, 2008 11:48 AM
> To: baetle
> Subject: Mapping between SIOC and EvoOnt, some other comments on EvoOnt
>
>
> I am looking at the latest version of the EvoOnt ontologies that can
> be found at [1]. I think there is potential for a more extensive
> mapping to SIOC.
>
> bom:Comment could be a subclass of sioc:Post

Agree

> bom:Issue seems to me like a subclass of sioc:Post, rather than
> sioc:Container.

I would rather say it could be a subclass of sioc:Thread because an Issue is
a discussion about a certain topic (defect). What do you think?

> bom:User is quite redundant with sioc:User. I think that rather than
> creating a new subclass, one could just add additional properties to
> sioc:User.

I accidentally redefined the imports. See below.

>
> Generally, I also suggest to add rdfs:labels (with a 'en' language
> tag) to all entities in EvoOnt, as many user interfaces depend on the
> existance of these labels. The ID part of URIs should not be used for
> human-readable labeling.

Sure. Will be in the next version.

>
> It also seems like the bug in the bug ontology still exists: bom:Issue
> is currently a subclass of two disjoint classes, which makes the
> ontology inconsistent.

The core of the problem is Protégé. As you might noticed, the ontologies are
generated using Protégé 4. At some point I must have accidentally set the
option "make siblings mutually disjoint" what caused Protégé to redefine all
the imported entities under the bom namespace. I'm currently "cleaning"
that out. The proper version validates without problems using pellet.
I will check the other ontologies and upload the latest version within the
next days.

Cheers
Jonas

Matthias Samwald

unread,
Oct 29, 2008, 9:12:10 AM10/29/08
to baetle

> I will check the other ontologies and upload the latest version within the
> next days.

Great! Thanks.

- Matthias

Jonas Tappolet

unread,
Oct 30, 2008, 6:14:21 PM10/30/08
to bae...@googlegroups.com
Hi

I uploaded the new versions.
- I tested the ontologies using pellet and the consistency bug should now
be fixed.
- The comments are now tagged with the @en language tag.
- Some more comments in the bug ontology added
- Issue is now a subclass of sioc:Thread

Again, you will find the ontologies under http://www.ifi.uzh.ch/ddis/evo

Cheers
Jonas

> -----Original Message-----
> From: bae...@googlegroups.com [mailto:bae...@googlegroups.com] On
> Behalf Of Matthias Samwald
> Sent: Wednesday, October 29, 2008 2:12 PM
> To: baetle
> Subject: Re: Mapping between SIOC and EvoOnt, some other comments on
> EvoOnt
>
>
>

Reply all
Reply to author
Forward
0 new messages