Comments on: Predicate Based Services http://www.ldodds.com/blog/2010/03/predicate-based-services/ A journal of no fixed aims or direction, by Leigh Dodds Fri, 10 Dec 2010 14:02:04 -0500 http://wordpress.org/?v=2.8.1 hourly 1 By: Semantic web (21/03/10) « pintiniblog http://www.ldodds.com/blog/2010/03/predicate-based-services/comment-page-1/#comment-423 Semantic web (21/03/10) « pintiniblog Sun, 21 Mar 2010 20:57:21 +0000 http://www.ldodds.com/blog/?p=448#comment-423 [...] Predicate Based Services (via Planet RDF, [...] [...] Predicate Based Services (via Planet RDF, [...]

]]> By: Jakob http://www.ldodds.com/blog/2010/03/predicate-based-services/comment-page-1/#comment-422 Jakob Thu, 18 Mar 2010 20:55:55 +0000 http://www.ldodds.com/blog/?p=448#comment-422 Do you know the <a href="http://jakoblog.de/2008/11/13/ariadne-article-about-seealso-linkserver-protocol/" rel="nofollow">SeeAlso linkserver API</a>? Given an identifier (as URI or as something like skos:prefLabel which can be mapped to its URI) you get a list of related resource. If you query for an identifier of resource $S you can get a list of ($O,$L) tuples with (in Turtle syntax) $S $P $O . $O rdfs:label $L . The predicate $P (let it be owl:sameAs or what you need) depends on the particular SeeAlso service but you can also put it in the "description" field of a SeeAlso response. The practical use case for SeeAlso is to provide links related to a given resource to show them in a library catalog. Instead of sameAs.org and similar services which are made for nerds that love RDF-triples, a SeeAlso service is made for normal web users first - they get a list of labeled (!) links and computers second. You can get a Perl implementation <a href="http://search.cpan.org/dist/SeeAlso-Server/" rel="nofollow">at CPAN</a> and a PHP preview implementation <a href="http://ws.gbv.de/beacon/phpbeacon.tgz" rel="nofollow">included in phpBeacon</a> (BEACON is a SeeAlso dump format that I work on at the moment). But you should also be able to hack the SeeAlso API in an hour or two its really simple. A citedBy-service based on ISBN numbers in the German Wikipedia is available at http://ws.gbv.de/seealso/isbn2wikipedia/ - it already supports RDF. Do you know the SeeAlso linkserver API? Given an identifier (as URI or as something like skos:prefLabel which can be mapped to its URI) you get a list of related resource. If you query for an identifier of resource $S you can get a list of ($O,$L) tuples with (in Turtle syntax)

$S $P $O . $O rdfs:label $L .

The predicate $P (let it be owl:sameAs or what you need) depends on the particular SeeAlso service but you can also put it in the “description” field of a SeeAlso response.

The practical use case for SeeAlso is to provide links related to a given resource to show them in a library catalog. Instead of sameAs.org and similar services which are made for nerds that love RDF-triples, a SeeAlso service is made for normal web users first – they get a list of labeled (!) links and computers second.

You can get a Perl implementation at CPAN and a PHP preview implementation included in phpBeacon (BEACON is a SeeAlso dump format that I work on at the moment). But you should also be able to hack the SeeAlso API in an hour or two its really simple.

A citedBy-service based on ISBN numbers in the German Wikipedia is available at http://ws.gbv.de/seealso/isbn2wikipedia/ – it already supports RDF.

]]> By: Hugh Glaser http://www.ldodds.com/blog/2010/03/predicate-based-services/comment-page-1/#comment-421 Hugh Glaser Wed, 10 Mar 2010 09:57:53 +0000 http://www.ldodds.com/blog/?p=448#comment-421 Thanks for the nice words Leigh. Good stuff. Sorry I haven't implemented yet :-) I think one of the really good variants of this would be narrow and broader, in the widest possible sense. So I am narrower than my family, but my nose is narrower than me. snafu is broader than fubar, etc..Clearly geographic stuff is relevant. If you don't get hung up on types, and just treat as a discovery mechanism, then it would give lots of leverage. Thanks for the nice words Leigh.

Good stuff.

Sorry I haven’t implemented yet :-)

I think one of the really good variants of this would be narrow and broader, in the widest possible sense. So I am narrower than my family, but my nose is narrower than me. snafu is broader than fubar, etc..Clearly geographic stuff is relevant.

If you don’t get hung up on types, and just treat as a discovery mechanism, then it would give lots of leverage.

]]> By: Most Tweeted Articles by Semantic Web Experts: MrTweet http://www.ldodds.com/blog/2010/03/predicate-based-services/comment-page-1/#comment-420 Most Tweeted Articles by Semantic Web Experts: MrTweet Tue, 09 Mar 2010 05:43:09 +0000 http://www.ldodds.com/blog/?p=448#comment-420 <strong>Your article was most tweeted by Semantic Web experts in the Twitterverse...</strong> Come see other top popular articles surfaced by Semantic Web experts!... Your article was most tweeted by Semantic Web experts in the Twitterverse…

Come see other top popular articles surfaced by Semantic Web experts!…

]]> By: admin http://www.ldodds.com/blog/2010/03/predicate-based-services/comment-page-1/#comment-418 admin Mon, 08 Mar 2010 12:01:22 +0000 http://www.ldodds.com/blog/?p=448#comment-418 Ed, Yes I think thats one useful way of looking at it. And I agree that Google is already essentially offering a service around vanilla hyperlinks. I think that one of the interesting things about this approach is that its fairly unique to the Linked Data world. There are lots of existing web services and applications that specialise in aggregating data about specific types of things (Photos, Blog Posts, Books) but very few that focus on relations. Thats mainly because without RDF & Linked Data those relations are not explicit. Google does a good job at sifting out potentially relevant links and relations, but we can do potentially do better. We may not be as comprehensive in all cases, but I think there's a reasonable trade-off there for some applications. Subj3ct seems to me more about trying to organize data around common URIs. I guess its similar in intent to sameAs.org, but doesn't quite fit my notion of the "topicOf" service. Ed,

Yes I think thats one useful way of looking at it. And I agree that Google is already essentially offering a service around vanilla hyperlinks.

I think that one of the interesting things about this approach is that its fairly unique to the Linked Data world. There are lots of existing web services and applications that specialise in aggregating data about specific types of things (Photos, Blog Posts, Books) but very few that focus on relations. Thats mainly because without RDF & Linked Data those relations are not explicit.

Google does a good job at sifting out potentially relevant links and relations, but we can do potentially do better. We may not be as comprehensive in all cases, but I think there’s a reasonable trade-off there for some applications.

Subj3ct seems to me more about trying to organize data around common URIs. I guess its similar in intent to sameAs.org, but doesn’t quite fit my notion of the “topicOf” service.

]]> By: Ed Summers http://www.ldodds.com/blog/2010/03/predicate-based-services/comment-page-1/#comment-417 Ed Summers Mon, 08 Mar 2010 11:21:41 +0000 http://www.ldodds.com/blog/?p=448#comment-417 I think it's a really neat idea Leigh. One way of thinking about these sameas-ish services is that they would basically be creating names (URIs) for useful and optimized queries of the web of data. One could argue that Google already offers a similar services around the hyperlink predicate (e.g. http://www.google.com/search?as_lq=http://www.ldodds.com). I also wonder how a service like https://subj3ct.com/ fits into your notion of topicOf? I think it’s a really neat idea Leigh. One way of thinking about these sameas-ish services is that they would basically be creating names (URIs) for useful and optimized queries of the web of data. One could argue that Google already offers a similar services around the hyperlink predicate (e.g. http://www.google.com/search?as_lq=http://www.ldodds.com). I also wonder how a service like https://subj3ct.com/ fits into your notion of topicOf?

]]> By: John S. Erickson, Ph.D. http://www.ldodds.com/blog/2010/03/predicate-based-services/comment-page-1/#comment-416 John S. Erickson, Ph.D. Mon, 08 Mar 2010 00:05:30 +0000 http://www.ldodds.com/blog/?p=448#comment-416 Thanks for the thought-provoking post, Leigh! My mind immediately turned to niche predicate- based services for publishing, media production and broadcast analytics (i.e. television ratings) domains. One facet of this would be a range of specializations on "madeBy," such as "shotBy," "editedBy," etc. I could also specialized services for certain scientific niches, esp. scientific workflow, genomics and pharma. An interesting evolution of these services would be the management of user "profiles" that would contextualize the lookups for users, in essence filtering the range of the results. One example that I think has actually been used before is restricting the range to certain CC licences. Interesting idea! Thanks for the thought-provoking post, Leigh!

My mind immediately turned to niche predicate- based services for publishing, media production and broadcast analytics (i.e. television ratings) domains. One facet of this would be a range of specializations on “madeBy,” such as “shotBy,” “editedBy,” etc.

I could also specialized services for certain scientific niches, esp. scientific workflow, genomics and pharma.

An interesting evolution of these services would be the management of user “profiles” that would contextualize the lookups for users, in essence filtering the range of the results. One example that I think has actually been used before is restricting the range to certain CC licences.

Interesting idea!

]]>