- From: Pat Hayes <phayes@ihmc.us>
- Date: Wed, 29 Jul 2009 14:41:08 -0500
- To: Simon Reinhardt <simon.reinhardt@koeln.de>
- Cc: public-lod@w3.org
On Jul 29, 2009, at 12:38 PM, Simon Reinhardt wrote: > Alan Ruttenberg wrote: >> On Wed, Jul 29, 2009 at 12:54 PM, Yves >> Raimond<yves.raimond@gmail.com> wrote: >>> and so we didn't got the incentive to >>> write a better one. Among those examples, you have: >>> >>> * A score in a musical performance >>> * A musical instrument in a musical performance >>> * A piece of text in a reading >>> * A microphone in a recording >> A chair in the room? The door to leave? The program handed out to the >> audience? The audience? The light bulb illuminating the room? The >> food >> that audience ate while watching? The videotape that was being used >> to >> record the performance? The city in which the performance took place? > > I think that's splitting hairs. If the light bulb is important to > you then add it to your data. With RDF it's always pretty much up to > you what you do, right? Wrong. At least in this sense. RDF doesn't itself constrain what is said, but RDF just allows the ontologies to be transmitted, and the whole point of an ontology like this is that it *does* constrain meanings. Why would I use a concept like event:Factor? Presumably so that others can (write software which will) understand what I intend to mean by using it. Its meaning, specified (perhaps loosely, but specified) by its home ontology, is one I wish to use in conveying my ideas by this usage. That is what such ontologies are for, it is their intended purpose. And if no specification is given at all, then there is no point in using it: I might as well make up my own URI and declare that it captures my intentions more exactly. And you, reading the RDF that I publish, will have no greater or lesser idea as to what it is that I intend to say than if I were to use event:Factor. Which is a reductio ad absurdum of the whole idea of having ontologies in the first place. > The ontology user and data publisher is as responsible for data > integration as is the ontology designer. Bears some responsibility, indeed, but the responsibility of the ontology user should be to use the ontology's terminology in ways that conform to the declared intent of the writer of the ontology. In order to exercise this responsibility, therefore, there must be at least some explanation of what that intention actually is. I do not ask for elaborate logical analysis, but surely it is not unreasonable to ask for something which is not absolutely vacuous. To be told that event:Factor means a factor of an event is to be told absolutely nothing. And to be told absolutely nothing in the cited documentation of an ontology being proposed for universal use is, in my world, somewhat offensive, even insulting. It amounts to reading a manual which says "This is so obvious that I cannot be bothered to even take the time to explain it to you." And this is not a good attitude to take towards the people who are, in effect, your customers. > And if the data consumers thing you went too far and have too much > noise in your data then you have to fix that. There is no way to know if data contains noise, when the concepts have no characterization. > And while the Event ontology doesn't state event:Factor and > geo:SpatialThing to be distinct (maybe they didn't want to make such > statements about other people's terms - with OWL 2 they could do > this for event:factor and event:place now though) I think it's > pretty obvious that you're supposed to use event:place for the city > in which the performance took place (or more exactly for the venue > which is in the city). Ah, if only such things were obvious. No, it is not obvious that event:place means the city. or even that one person's place might not be another person's factor. And BTW, your mention of venue raises an interesting question: can one event be in several event:places? For example, if the recording was made in a studio in Los Angeles, would it be acceptable to have the event:place be USA? Or would that simply be wrong? The point being, of course, that shouldn't the ontology specification say **something** about issues like this? Questions like this matter when one wants to retrieve data based on conditions (find all the recordings that were made anywhere in the USA before 1997.) Pat Hayes > > Regards, > Simon > > > ------------------------------------------------------------ IHMC (850)434 8903 or (650)494 3973 40 South Alcaniz St. (850)202 4416 office Pensacola (850)202 4440 fax FL 32502 (850)291 0667 mobile phayesAT-SIGNihmc.us http://www.ihmc.us/users/phayes
Received on Wednesday, 29 July 2009 19:41:48 UTC