The Similarity Ontology is discussed in When owl:sameAs isn’t the Same: An Analysis of Identity in Linked Data. It defines properties that can be used when you almost mean owl:sameAs.

The properties seem useful. However, it isn't clear how actively maintained or used the ontology is. For example, a link to a supporting ontology defined in the same paper doesn't work. The preferred prefix isn't registered in prefix.cc. And the ontology itself makes assertions that don't match the human-readable comments.

For example, the "matches" property is defined with the following rdf:types
<owl:ObjectProperty rdf:about="#matches"> <rdf:type rdf:resource="&owl;SymmetricProperty"/> <rdf:type rdf:resource="&owl;TransitiveProperty"/>

Which I would assume make it symmetric and transitive, but not reflexive. However, the comment says "A is a match to B, they are exactly alike, but may not be one in the same. This property is reflexive and symmetric."

Does anyone know whether this ontology is used in practice and what its maintenance status is?

asked 07 Jan '13, 14:58

Lin%20Clark's gravatar image

Lin Clark
31717
accept rate: 0%

edited 07 Jan '13, 14:59


Another option to see how often some property or class is used is vocab.cc, which I believe indexes content from the Billion Triple Challenge dataset(s). The system is a little broken in that it returns an error instead of no results, and it seems to return an error for many of those properties. I don't believe that the Similarity Ontology has seen much by way of adoption, no.

In any case, the semantics of the properties in that ontology are not very useful and are by no means an alternative for owl:sameAs. The semantics of owl:sameAs are given in the OWL standard and support the key principle of replacement (any identifier can be replaced by an "equivalent" identifier anywhere it is used). The Similarity Ontology doesn't/can't support this principle ... it only supports making links transitive or symmetric or reflexive or some combination thereof.

The only complementary option I see that's needed for owl:sameAs is a non-symmetric version that applies replacement in one direction. Say you own a small rental store and have a movie for rent ... we can be topical and say Django Unchained. The IMDb page has some OGP RDFa you can re-use (though not ideal, sufficient for this example). The movie in the context of your store has a rental price. The movie in the context of IMDb doesn't have a rental price. What you want to do is "import" information from IMDb into your local rental store site (but not vice-versa) ... you want a non-symmetric version of owl:sameAs that applies replacement only in the direction of the IMDb movie into your local movie resource. This way, at worst, you're only screwing up your local resource. The availability of such a property could also lead to prevalent use of a Prototype pattern, which I think would work wonders on the Web of Data. (For this "owl:specialises" property to become a reality, the OWL guys need to write the semantics of such a property into the specification. I wouldn't hold my breath.)

With respect to the paper in question, though it raises an important issue and has interesting discussion/summarisation of the problem, I don't believe that the problem is quite as widespread as their analysis makes out to be. Sure, there's some really bad owl:sameAs use in the Wild, but my own experience is such that it's not half as bad as they make it out to be. Drug data is very broken and occasionally you'll see things like music artists muddled up with their albums through a series of owl:sameAs relations across DBpedia and YAGO, but in my experience (a recent example of a prototype which, in the background, applies owl:sameAs semantics to smush together the entire BTC 2012 dataset in as naive a manner as you can get without the world ending), owl:sameAs and blank-nodes are the new straw-men of the Semantic Web and Linked Data. There are much bigger problems on our hands.

I'll take a Web of Data with owl:sameAs and its strong semantics and its occasional flaws over a Web of Data without owl:sameAs or with weak alternatives any day of the week.

link

answered 07 Jan '13, 20:16

Signified's gravatar image

Signified ♦
23.5k1623
accept rate: 37%

edited 07 Jan '13, 20:20

link

answered 07 Jan '13, 18:36

Sweet%20Burlap's gravatar image

Sweet Burlap
2.7k38
accept rate: 18%

Your answer
toggle preview

Follow this question

By Email:

Once you sign in you will be able to subscribe for any updates here

By RSS:

Answers

Answers and Comments

Markdown Basics

  • *italic* or _italic_
  • **bold** or __bold__
  • link:[text](http://url.com/ "Title")
  • image?![alt text](/path/img.jpg "Title")
  • numbered list: 1. Foo 2. Bar
  • to add a line break simply add two spaces to where you would like the new line to be.
  • basic HTML tags are also supported

Tags:

×15
×9

Asked: 07 Jan '13, 14:58

Seen: 572 times

Last updated: 07 Jan '13, 20:20