Comments on: When Linked Data Rules Fail http://www.mkbergman.com/846/when-linked-data-rules-fail/ Mike Bergman on the semantic Web and structured Web Mon, 06 Feb 2012 00:15:33 +0000 hourly 1 http://wordpress.org/?v=3.2.1 By: The next web to be user-centric (thoughts on David Siegel’s Pull book) « GrowthTimes http://www.mkbergman.com/846/when-linked-data-rules-fail/comment-page-1/#comment-92293 The next web to be user-centric (thoughts on David Siegel’s Pull book) « GrowthTimes Thu, 09 Jun 2011 19:03:47 +0000 http://www.mkbergman.com/?p=846#comment-92293 [...] When Linked Data Rules Fail (mkbergman.com) [...] [...] When Linked Data Rules Fail (mkbergman.com) [...]

]]> By: The Nature of Connectedness on the Web » AI3:::Adaptive Information http://www.mkbergman.com/846/when-linked-data-rules-fail/comment-page-1/#comment-62427 The Nature of Connectedness on the Web » AI3:::Adaptive Information Mon, 22 Nov 2010 14:43:00 +0000 http://www.mkbergman.com/?p=846#comment-62427 [...] example is one that Fred Giasson and I brought up one year ago in When Linked Data Rules Fail [2]. That piece discussed many poor practices within linked data, and used as one case the [...] [...] example is one that Fred Giasson and I brought up one year ago in When Linked Data Rules Fail [2]. That piece discussed many poor practices within linked data, and used as one case the [...]

]]> By: Bill Roberts http://www.mkbergman.com/846/when-linked-data-rules-fail/comment-page-1/#comment-52755 Bill Roberts Thu, 19 Nov 2009 10:38:35 +0000 http://www.mkbergman.com/?p=846#comment-52755 Thanks for this in-depth analysis. While I think it's great that TWI and NYT have gone ahead to publish their data, it's important to look at how linked data could be done better. A few weeks ago I asked on my blog <a href="http://www.webofdatablog.com/articles/2009/10/20/what-makes-good-linked-data" rel="nofollow">What Makes Good Linked Data?</a> Your discussion above goes a long way to helping answer that question! The NYT case seems to be a common difficulty that people have with representing info in RDF: distinguishing between non-information and information resources, between the person and articles about that person. I'm not sure how we help people get to grips with that issue. And with TWI and data.gov, I have a lot of respect for the work of the guys behind that project, I'm sure they understand the issues - so I hope that the work so far is a first step and they will go further to surface the true semantics of the data. Clearly that involves a lot more work than the current naïve direct table column to predicate mappings. I can understand the reason that people want to shout about "data.gov now available as linked data", but as you point out, maybe that is a bit premature. Thanks for this in-depth analysis. While I think it’s great that TWI and NYT have gone ahead to publish their data, it’s important to look at how linked data could be done better. A few weeks ago I asked on my blog What Makes Good Linked Data? Your discussion above goes a long way to helping answer that question!

The NYT case seems to be a common difficulty that people have with representing info in RDF: distinguishing between non-information and information resources, between the person and articles about that person. I’m not sure how we help people get to grips with that issue.

And with TWI and data.gov, I have a lot of respect for the work of the guys behind that project, I’m sure they understand the issues – so I hope that the work so far is a first step and they will go further to surface the true semantics of the data. Clearly that involves a lot more work than the current naïve direct table column to predicate mappings. I can understand the reason that people want to shout about “data.gov now available as linked data”, but as you point out, maybe that is a bit premature.

]]> By: Ben http://www.mkbergman.com/846/when-linked-data-rules-fail/comment-page-1/#comment-52748 Ben Thu, 19 Nov 2009 00:10:57 +0000 http://www.mkbergman.com/?p=846#comment-52748 Thanks for the important post - though I have to concur with Denny's last comment there (#11). I don't think that these efforts are completely without value. <strong> If the NYT et al hadn't put that RDF out there they wouldn't have enjoyed the benefit of your freely provided semantic analysis ;)</strong>. This is a debate I often get into with my old advisor Mark Wilkinson. He would agree with you completely. While I share his and your frustration with the missed opportunities that occur when things on the semantic web are left un/improperly defined, I still have to lend my support to the groups that start their work by 'just putting it out there' - I just hope they are listening to responses like this and have a plan for iterative improvement. (BTW, the resolution of a <a href="http://i9606.blogspot.com/2009/02/semantic-dissonance-in-uniprot.html" rel="nofollow">very similar problem I discussed a while ago in the UniProt database</a> was apparently to drop the use of owl:sameAs in favor of seeAlso - not wrong I guess, but still dissappointing.) Thanks for the important post – though I have to concur with Denny’s last comment there (#11). I don’t think that these efforts are completely without value. If the NYT et al hadn’t put that RDF out there they wouldn’t have enjoyed the benefit of your freely provided semantic analysis ;) . This is a debate I often get into with my old advisor Mark Wilkinson. He would agree with you completely. While I share his and your frustration with the missed opportunities that occur when things on the semantic web are left un/improperly defined, I still have to lend my support to the groups that start their work by ‘just putting it out there’ – I just hope they are listening to responses like this and have a plan for iterative improvement. (BTW, the resolution of a very similar problem I discussed a while ago in the UniProt database was apparently to drop the use of owl:sameAs in favor of seeAlso – not wrong I guess, but still dissappointing.)

]]>