Re: tweet2rdf vocabulary convergence

Hi,

In addition to Uldis' reply, some more thoughts on reusing existing  
vocabularies

On 28 Sep 2009, at 12:46, Benjamin Nowack wrote:

> On 28.09.2009 11:41:06, Michael Hausenblas wrote:
> [...]
>> I'd be very interested to engage in this. Please let me know if you  
>> have
>> concrete steps planned. Maybe first review existing vocabs to see  
>> what is
>> covered already, today?
> I think a lot is available already, so what we mainly need is probably
> mainly a "suggested vocabulary blend":

+1

> * type
>      * rss:item
>      * sioc:MicroblogPost
> * all typical rss properties
>      * rss:title
>      * rss:description
>      * content:encoded
>      * dc:subject (hashtags)

One issue with dc:subject is that is strictly refers to something  
which is the "subject" of the post, not only a topic.
E.g. if I say "#sunny weather in Galway" , "sunny" is probably not the  
subject of my tweet, just something I want to emphasize, see also [1]
(this is one reason I introduced a new moat:taggedWith property in  
MOAT [2])

>      * dc:creator for the author name

What about foaf:maker / sioc:has_creator in addition here, in order to  
provide interlinking rather than just plain-text author information

> * author stuff
>      * dct:creator for author resource (profile vs person issue?)
>      * sioc:has_creator for author profile/account URL
>      * sioc:avatar / atom:link
> * links
>      * dct:references (short or expanded? prolly the latter)
> * Semweb URIs
>      * dct:subject
> * threads
>      * sioc:reply_of
>
> What's missing (not sure which are really useful or possible):
> * type
>      * ex:DirectMessage
>      * ex:ReTweet
> * targetUser (@user)
> * mentioned user (... @user)
> * referred by ( ... via @user)
> * machine/triple tags?
> * derived stats?
>      * number of re-tweets
>      * number of url posts

There are num_replies / num_views in SIOC that can be used for such  
purposes.

> * client app (needed? there's twitter:source, but has html)

I'd suggest "admin:generatorAgent" here [3], as in some FOAF-builders  
such as foaf-o-matic [4]

> * ratings (if so, which notation?)

What about the review vocabulary and its review:rating property [5] ?
A question might be to find mappings between user syntax and such  
reviews

> * group (identi.ca uses !group, but doesn't have rdf, I think)
>

Might sioc:UserGroup or foaf:Group be used here for groups of users /  
people ?
If the group needs to support not only people but also tweets, topics,  
etc. sioc:Community can be well-suited

Hope that helps,

Alex.

[1] http://librarytechnz.natlib.govt.nz/2008/09/adding-tags-to-dc-metadata.html
[2] http://moat-project.org/ns
[3] http://webns.net/mvcb/
[4] http://www.ldodds.com/foaf/foaf-a-matic
[5] http://vocab.org/review/


> Benji
>
> --
> Benjamin Nowack
> http://bnode.org/
> http://semsol.com/
>
>
>>
>> Cheers,
>>     Michael
>>
>> --
>> Dr. Michael Hausenblas
>> LiDRC - Linked Data Research Centre
>> DERI - Digital Enterprise Research Institute
>> NUIG - National University of Ireland, Galway
>> Ireland, Europe
>> Tel. +353 91 495730
>> http://linkeddata.deri.ie/
>> http://sw-app.org/about.html
>>
>>
>>
>>> From: Benjamin Nowack <bnowack@semsol.com>
>>> Organization: semsol.com
>>> Reply-To: Benjamin Nowack <bnowack@semsol.com>
>>> Date: Mon, 28 Sep 2009 11:35:53 +0200
>>> To: Semantic Web community <semantic-web@w3.org>
>>> Subject: tweet2rdf vocabulary convergence
>>> Resent-From: Semantic Web community <semantic-web@w3.org>
>>> Resent-Date: Mon, 28 Sep 2009 09:36:31 +0000
>>>
>>>
>>> Hi,
>>>
>>> Morton Swimmer suggested that there might be broader interest to  
>>> talk
>>> a bit about RDF extracted from tweets, so here we go:
>>>
>>> There are multiple tools and services that convert twitter profiles
>>> and contacts to RDF (e.g semantictweet[1] or knowee), I think they  
>>> all
>>> mostly re-use stuff from FOAF and don't really need new terms.
>>>
>>> But there are also tools that convert individual tweets to RDF
>>> (I think Tom Morris had code. smesher is another example), or the
>>> other way round (e.g. SMOB). Streams can nicely be grounded in RSS,
>>> possibly with an additional sioc:MicroblogPost type, but what about
>>> the semi-structured data? Should we try to create a shared vocab for
>>> such in-tweet data (recipient, mentioned people, author-avatar/ 
>>> profile,
>>> tags, machine tags, short urls, expanded urls, re-tweets, vias,
>>> embedded Linked Data URIs, groups, DM, ...)?
>>>
>>> I've been playing a bit with in-tweet structures[2] a while ago, but
>>> so far mainly made up app-specific terms. For a new project, I'm
>>> extracting ratings and moods (via evolving patterns similar to
>>> nanoformats [3], twitterdata[4], or simple word lists). I'm again
>>> making up one-off terms here, too, and could surely benefit from a
>>> more stable vocab.
>>>
>>> Anyone interested in exploring this a little further? VoCamp near
>>> Düsseldorf or Amsterdam, maybe? ;)
>>>
>>> Cheers,
>>> Benji
>>>
>>>
>>> [1] http://semantictweet.com/
>>> [2] http://www.smesher.org/media/2009/02/13/SMR_RDFExtractor.phps
>>> [3] http://microformats.org/wiki/microblogging-nanoformats
>>> [4] http://twitterdata.org/
>>>
>>> --
>>> Benjamin Nowack
>>> http://bnode.org/
>>> http://semsol.com/
>>>
>>>
>>
>
>

--
Dr. Alexandre Passant
Digital Enterprise Research Institute
National University of Ireland, Galway
:me owl:sameAs <http://apassant.net/alex> .

Received on Monday, 28 September 2009 12:57:09 UTC