Comments on: The trouble with PSI’s http://www.marcdegraauw.com/2007/01/28/the-trouble-with-psis/ Some thoughts on meaning, the web and everything Fri, 06 Aug 2010 16:01:38 +0000 hourly 1 http://wordpress.org/?v=3.3.1 By: David Byrden http://www.marcdegraauw.com/2007/01/28/the-trouble-with-psis/#comment-9182 David Byrden Fri, 25 Jul 2008 13:48:30 +0000 http://www.marcdegraauw.com/2007/01/28/the-trouble-with-psis/#comment-9182 >> "PSI’s are URI’s which uniquely identify something" To be more correct, PSIs are the resources that you get when you resolve the URIs. >> “PSI’s are URI’s which uniquely identify something”

To be more correct, PSIs are the resources that you get when you resolve the URIs.

]]> By: chris sizemore http://www.marcdegraauw.com/2007/01/28/the-trouble-with-psis/#comment-6305 chris sizemore Thu, 28 Feb 2008 22:51:08 +0000 http://www.marcdegraauw.com/2007/01/28/the-trouble-with-psis/#comment-6305 make it easy on yourself and just use Wikipedia URIs (or, if you must, dbPedia.org URIs)... you are both worrying too much... ;-) make it easy on yourself and just use Wikipedia URIs (or, if you must, dbPedia.org URIs)…

you are both worrying too much… ;-)

]]> By: Lars Marius http://www.marcdegraauw.com/2007/01/28/the-trouble-with-psis/#comment-177 Lars Marius Mon, 29 Jan 2007 11:58:06 +0000 http://www.marcdegraauw.com/2007/01/28/the-trouble-with-psis/#comment-177 Frankly, I think the biggest problem with PSIs is that they've been oversold, so that lots of people think they are something they're not, and others dismiss them for failing to do what they were never intended to do. I'll try to put down what I think of your objections. In general, I think they are valid and that it's good that you point them out, because that helps dispel all the confusion around this. I think you overstate the case a bit (e.g: "PSIs solve the wrong problem"), but I think that's mostly because of all the overblown PSI marketing. With luck we'll find that after the stick has been pushed too far first one way and then the other it will finally be sticking straight up. :-) 1: Yes, the pollution problem is real. No, PSIs will not solve it, or even affect it at all. That just means we have a challenge, though. It's not a reason to give up on PSIs, just as governments don't give up on social security numbers because of pollution problems. 2: You are right that the Semagia PSI page provides no information beyond what knowing that a date is an ISO 8601 date will. But then that's a special case. The PSI for, say tm:supertype-subtype will (once published) define the meaning of that association type. A URI can't do that. So for subjects where there actually is something useful to define, the subject indicator will do that. Where there isn't something useful to define PSIs add nothing to basic URIs, and you might as well skip the indicator. 3: It's definitely true that getting everyone to use the same PSI for every concept is a hopeless task, but I think your cpr:person example shows perfectly what you can do with URIs to identify concepts. With global identifiers (PSIs or just plain URIs) you can reuse someone else's concept where that works, or relate your own concept to someone else's (with subclassing, DL expressions, or whatever). Frankly, I think the biggest problem with PSIs is that they’ve been oversold, so that lots of people think they are something they’re not, and others dismiss them for failing to do what they were never intended to do.

I’ll try to put down what I think of your objections. In general, I think they are valid and that it’s good that you point them out, because that helps dispel all the confusion around this. I think you overstate the case a bit (e.g: “PSIs solve the wrong problem”), but I think that’s mostly because of all the overblown PSI marketing. With luck we’ll find that after the stick has been pushed too far first one way and then the other it will finally be sticking straight up. :-)

1: Yes, the pollution problem is real. No, PSIs will not solve it, or even affect it at all. That just means we have a challenge, though. It’s not a reason to give up on PSIs, just as governments don’t give up on social security numbers because of pollution problems.

2: You are right that the Semagia PSI page provides no information beyond what knowing that a date is an ISO 8601 date will. But then that’s a special case. The PSI for, say tm:supertype-subtype will (once published) define the meaning of that association type. A URI can’t do that. So for subjects where there actually is something useful to define, the subject indicator will do that. Where there isn’t something useful to define PSIs add nothing to basic URIs, and you might as well skip the indicator.

3: It’s definitely true that getting everyone to use the same PSI for every concept is a hopeless task, but I think your cpr:person example shows perfectly what you can do with URIs to identify concepts. With global identifiers (PSIs or just plain URIs) you can reuse someone else’s concept where that works, or relate your own concept to someone else’s (with subclassing, DL expressions, or whatever).

]]>