The semantic ambiguity of linked data has been a hot topic of discussion of late.
Not only of what linked data links to but of linked data itself!
If you have invested a lot in linked data efforts, don’t panic!
Topic maps, even using XTM/CTM syntaxes, to say nothing of more exotic models, can reduce any semantic ambiguity using occurrences.
If and when it is necessary.
Quite serious, “if and when necessary.”
Err, “if and when necessary” meaning when it is important enough for someone to pay for the disambiguation.
Ambiguity between buyers and sellers of women’s shoes or lingerie probably abounds, but unless someone is willing to pay the freight for disambiguation, it isn’t my concern.
Linked data is exposing the ambiguity of the Semantic Web.
Being unable to solve the semantic ambiguity it exposes, linked data is creating opportunities for topic maps!
Maybe we should send the W3C a fruit basket or something?
[…] This post was mentioned on Twitter by Semantic Web Blogs, IT Blog Network. IT Blog Network said: #Semantic #Blogs LOD, Semantic Ambiguity and Topic Maps « Another Word For It: Linked data is exposing the ambig… http://bit.ly/dgwa1H […]
Pingback by Tweets that mention LOD, Semantic Ambiguity and Topic Maps « Another Word For It -- Topsy.com — November 12, 2010 @ 7:54 pm
Please provide a link to a resource that demonstrates how Topic Map addresses ambiguity in a manner that’s unachievable using Linked Data.
Your point would be easier to understand, and ultimately much more credible with a link to a live example, that supports your claim.
I await your link.
Kingsley
Comment by Kingsley Idehen — November 13, 2010 @ 12:49 pm
(link below)
Your question would be easier to answer if you did not use marketing department claims like “unachievable using Linked Data.”
What does “unachievable using Linked Data” mean?
Does it mean:
1) Following the published Linked Data standards?
2) #1 plus some variety of 303, 200* or other non-standard solution to URI ambiguity?
3) #2 plus just believe that Linked Data works and it will be alright?
Topic maps routinely disambiguate between URIs that are identifiers and those that are resources, without the use of #1, #2 or #3 above.
For a classic example of a topic map, see: http://www.ontopia.net/operamap/index.jsp
Topic maps in production use, such as by the Norwegian post office, make such distinctions.
If I understood why software has any relationship to understanding the nature of the problem to be addressed I might be able to give a more helpful answer.
Comment by Patrick Durusau — November 14, 2010 @ 7:21 am
Don’t know what you mean by marketing terminology. Why not just focus on the matter at hand? Don’t use name calling to detract from a blatant challenge. I am challenging you to back up your rhetoric!
What is this: http://www.ontopia.net/operamap/index.jsp ? What is: http://www.ontopia.net/operamap/theatre.jsp?id=cairo-opera ?
Are either one of those examples of Subject and Resource disambiguation? Meaning: a machine or a human can unambiguously discern the Document that Describes an Entity, and the Entity itself?
Hopefully, the paragraph above isn’t marketing rhetoric. Also, is HTML the only representation you serve? I ask because of the output below:
1. http://redbot.org/?uri=http%3A%2F%2Fwww.ontopia.net%2Foperamap%2Ftheatre.jsp%3Fid%3Dcairo-opera — how does a machine know if you have alternative data representations re. resource content? Nothing in HTTP response headers, nothing in section of the HTML
Now on the Linked Data side of things look at this for clarity:
1. http://redbot.org/?uri=http%3A%2F%2Fdbpedia.org%2Fdescribe%2F%3Furi%3Dhttp%3A%2F%2Fdbpedia.org%2Fresource%2FEgypt — Notice rich HTTP response metadata, also note includes matching metadata
2. http://dbpedia.org/describe/?uri=http://dbpedia.org/resource/Egypt — HTML Resource .
I also encourage you to digest:
1. http://lab.linkeddata.deri.ie/2010/star-scheme-by-example/ — at the very least, show me a Topic Maps variant of this should a Resource exist.
Comment by Kingsley Idehen — November 14, 2010 @ 1:50 pm
I asked you to define what you meant by unachievable by Linked Data?
Since you skipped that part of my post, I assume you have no answer for it.
The issue that seems elusive in this conversation is that there is no general solution to the ambiguity problem.
What the topic map example illustrates is distinguishing between URIs for identifiers and URIs that are pointers to resources. That is one form of ambiguity.
My point being that any solution has to choose and address some portion of ambiguity and there is no general solution that does it all.
Since there is no working software to demonstrate the Incompleteness Theorem I assume that lacks credibility as well?
BTW, your “Star Schema By Example” document has a fatal flaw: It has section titles of “costs and benefits” but only asserts the “benefits” and not the “costs.” Or for that matter any of the problems such as ambiguity.
I would not show that document to anyone until it gets fixed.
Comment by Patrick Durusau — November 14, 2010 @ 8:58 pm