To: | "[ontolog-forum]" <ontolog-forum@xxxxxxxxxxxxxxxx> |
---|---|
From: | Amanda Vizedom <amanda.vizedom@xxxxxxxxx> |
Date: | Fri, 15 Mar 2013 16:38:44 -0400 |
Message-id: | <CAEmngXvm85Aq45F7bX5P6TSD4rhVQ-DgEOJTt_BAVQYgzv635A@xxxxxxxxxxxxxx> |
Indeed. I agree with both Leo and Simon here, and want to emphasize their points; this is a very common situation and there are corresponding, also very common mistakes, in re-use.
In particular:
1. NAICS is presented as a taxonomy but in fact does not have consistent taxonomic semantics. 2. In particular, the parent/child relationship does not have consistent semantics. Another way to put this is that it is subject to "semantic drift" -- it may start out as a subclass relationship at one level, change to something partitive further down. I don't recall how widely this drift ranges for NAICS, but it is very common to see drift across a handful of different parent/child relationships within a single "taxonomy" or "classification system". Besides subclass and partOf (and, indeed, different subProperties of partOf), it's common to find such relations as subtopic, instanceOf, ownedBy, locatedIn, and variants of controlledBy all mixed into what appears to be a single hierarchy.
3. This has particular consequence for automated reuse in ontological form. In the best cases, the semantic drift is predictable by level. If there is some pattern similar to what Simon attempts to identify, then at least a script or tool can utilize the level to determine what parent/child relationship to assert. That's the best case. In other cases there is no consistent pattern like this, or at least none that can be readily identifiable. That means that tools such as the TBC spreadsheet importer can turn the artifact into syntactic OWL, but only the most undefined relationship, something like BT/NT, can be asserted automatically. If subClass or any of the other more meaningful relations is used, much of the resulting OWL will wrong.
4. However, the existence of such artifacts is often a sign that the system in question would be better modeled with an ontology. If the developers had a need to capture such a variety of hierarchical relationships, it is very likely that having those relationships made explicit would be of benefit to at least some users. Most ways of using the artifact for automated information processing will be undermined by the inconsistency.
5. These observations brought to you by quite a few experiences with the process and/or results of turning users' "taxonomies" and "ontologies," as well as expert-developed and widely used "taxonomies" and "ontologies" in users' domains (such as NAICS, MeSH, CheBI, and many others) into usable ontologies for application purposes. It's often worth it, but the transformation to explicit semantics can be non-trivial (and require verification of the resulting model to make sure that the varying semantics have been interpreted correctly.
Best, Amanda On Fri, Mar 15, 2013 at 3:57 PM, Obrst, Leo J. <lobrst@xxxxxxxxx> wrote:
_________________________________________________________________ Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/ Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/ Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx Shared Files: http://ontolog.cim3.net/file/ Community Wiki: http://ontolog.cim3.net/wiki/ To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J (01) |
Previous by Date: | Re: [ontolog-forum] Fwd: MOVED: Re: [ontology-summit] Hackathon: BACnet Ontology, Obrst, Leo J. |
---|---|
Next by Date: | Re: [ontolog-forum] owl:sameAs - Is it used in a right way?, Rich Cooper |
Previous by Thread: | Re: [ontolog-forum] Fwd: MOVED: Re: [ontology-summit] Hackathon: BACnet Ontology, Obrst, Leo J. |
Next by Thread: | Re: [ontolog-forum] Fwd: MOVED: Re: [ontology-summit] Hackathon: BACnet Ontology, Obrst, Leo J. |
Indexes: | [Date] [Thread] [Top] [All Lists] |