Comments on: Announcing UMBEL: A Lightweight Subject Structure for the Web http://www.mkbergman.com/387/announcing-umbel-a-lightweight-subject-structure-for-the-web/ Mike Bergman on the semantic Web and structured Web Wed, 01 Feb 2012 21:05:31 +0000 hourly 1 http://wordpress.org/?v=3.2.1 By: Danny http://www.mkbergman.com/387/announcing-umbel-a-lightweight-subject-structure-for-the-web/comment-page-1/#comment-45622 Danny Sun, 15 Jul 2007 07:41:54 +0000 http://www.mkbergman.com/?p=387#comment-45622 Hi Mike, it sounds interesting and potentially very useful. Can't resist offering an alternate <a href="http://en.wikipedia.org/wiki/Humble_pie" rel="nofollow">etymology</a>, I mean, the project does seem to be about creating some kind of internal organs & piping... Hi Mike, it sounds interesting and potentially very useful.

Can’t resist offering an alternate etymology, I mean, the project does seem to be about creating some kind of internal organs & piping…

]]> By: Mike http://www.mkbergman.com/387/announcing-umbel-a-lightweight-subject-structure-for-the-web/comment-page-1/#comment-45598 Mike Fri, 13 Jul 2007 14:41:24 +0000 http://www.mkbergman.com/?p=387#comment-45598 Hi Lars, You are correct that UMBEL is similar to dc:subject or foaf:interest (there are others). However, unlike those approaches, there will be a reference set of subject 'proxy' names to pick from and synonym-like relationships similar to WordNet synsets. As intended, a single proxy name can therefore relate to the embracing "concept" of the subject without getting overly hung up on the precise name or description. Section 3.1 of the draft UMBEL specification, in fact, touches a bit on the use of UMBEL in relation to such other systems: "In addition to their use as a binding layer, this standard listing of subjects can also be referenced by resources described by other ontologies (e.g., dc:subject or foaf:interest)." You are also correct on UMBEL's intent as a means to describe the subject(s) of various data sets, applicable to Topic Maps and other data formalisms, so that others can find it. Finally, RDF was chosen as the internal representation of UMBEL over Topic Maps for a number of reasons: 1) RDF appears to be a natural "middle ground" in the spectrum of data formalisms (see my <a href="http://www.mkbergman.com/?p=374" rel="nofollow">earlier post</a> on this); 2) SKOS, the actual intended language for UMBEL, is a RDF Schema designed for the kind of information and classification purpose of UMBEL; and 3) there are RDF/SKOS-Topic Map interoperability prospects that <a href="http://www.garshol.priv.no/blog/10.html" rel="nofollow">you</a> have described directly and in support of the <a href="http://www.w3.org/TR/rdftm-survey/" rel="nofollow">W3C</a>. Most importantly, the choice of RDF (SKOS, actually) is really meant to be an internal representation and not prejudicial to the use of UMBEL in relation to any other data formalism. The whole point is to be neutral (as much as possible! :) ) and applicable to a number of important frameworks. As someone so involved with Topic Maps over the years, I hope you can keep an open mind at minimum and preferably a helping hand in this effort. I encourage you to share your thoughts and concerns on the UMBEL Google forum. Regards, Mike Hi Lars,

You are correct that UMBEL is similar to dc:subject or foaf:interest (there are others). However, unlike those approaches, there will be a reference set of subject ‘proxy’ names to pick from and synonym-like relationships similar to WordNet synsets. As intended, a single proxy name can therefore relate to the embracing “concept” of the subject without getting overly hung up on the precise name or description. Section 3.1 of the draft UMBEL specification, in fact, touches a bit on the use of UMBEL in relation to such other systems: “In addition to their use as a binding layer, this standard listing of subjects can also be referenced by resources described by other ontologies (e.g., dc:subject or foaf:interest).”

You are also correct on UMBEL’s intent as a means to describe the subject(s) of various data sets, applicable to Topic Maps and other data formalisms, so that others can find it.

Finally, RDF was chosen as the internal representation of UMBEL over Topic Maps for a number of reasons: 1) RDF appears to be a natural “middle ground” in the spectrum of data formalisms (see my earlier post on this); 2) SKOS, the actual intended language for UMBEL, is a RDF Schema designed for the kind of information and classification purpose of UMBEL; and 3) there are RDF/SKOS-Topic Map interoperability prospects that you have described directly and in support of the W3C. Most importantly, the choice of RDF (SKOS, actually) is really meant to be an internal representation and not prejudicial to the use of UMBEL in relation to any other data formalism.

The whole point is to be neutral (as much as possible! :) ) and applicable to a number of important frameworks. As someone so involved with Topic Maps over the years, I hope you can keep an open mind at minimum and preferably a helping hand in this effort.

I encourage you to share your thoughts and concerns on the UMBEL Google forum.

Regards, Mike

]]> By: Lars Marius Garshol http://www.mkbergman.com/387/announcing-umbel-a-lightweight-subject-structure-for-the-web/comment-page-1/#comment-45595 Lars Marius Garshol Fri, 13 Jul 2007 11:21:55 +0000 http://www.mkbergman.com/?p=387#comment-45595 Your "isAbout" relationship sounds very much like dc:subject from Dublin Core. In fact, they are probably the same relationship. It might be useful for UMBEL to make the relationship between these two relationships clear. I'm not 100% clear on what UMBEL is. Is it intended to, for example, allow me to register my Topic Maps PSI set (could be an ontology (that is, just a set of entity and property types), or a taxonomy, or some other kind of data structure) so others can find it? If so, why RDF, and why not Topic Maps, given that inferencing was not a key requirement? Your “isAbout” relationship sounds very much like dc:subject from Dublin Core. In fact, they are probably the same relationship. It might be useful for UMBEL to make the relationship between these two relationships clear.

I’m not 100% clear on what UMBEL is. Is it intended to, for example, allow me to register my Topic Maps PSI set (could be an ontology (that is, just a set of entity and property types), or a taxonomy, or some other kind of data structure) so others can find it?

If so, why RDF, and why not Topic Maps, given that inferencing was not a key requirement?

]]>