About the model: Question 12
12. A related question. FRAD also treats key identifier (unique machine-readable identifier for an entity) as a class rather than a property. Why not treat key identifier as a property of person?
“Indeed, why not? I don’t know of any standard that has tried to solvepersistent identification as well as Topic Maps. Here’s what we do ; Topic 234 [person] [psi] “http://psi.marktwainfoundantion.org/mark_twain” name “Mark Twain” name [alternate] “Mr Twain” We use Public Subject Indicators [psi] as properties which, really,are URI’s that you have published and are committed to keep alive,meaningful and persistent. If you are committed to keeping the PSI for“Martha Yee” alive and well, you publish your PSI and make sure thatdoesn’t resolve to anything 40x (meaning, not found, unless that’sintentional). These are basically global constants, and they (as thename suggest) indicate subjects. Institutions and organisations mapwhat they want, but as long as they’re PSI’s they are committed tothose URI’s, making sure that if *others* out there use them they willbe useful in the future as well. Of course, this doesn’t stop you from making up imaginary URI’s oreven existent but not really in your domain; it’s crazy to think thatall organizations of the world will commit to PSI’s, little less knowabout Topic Maps or persistent identifiers, so some leeway must begiven. But at least it’s a mechanism that works well. There are whole sections of the Topic Maps standard that deal withPSI’s and rules for merging Topic Maps together with them. This way,as long as we talk about the same PSI’s we can merge smaller anddiverse Topic Maps together at any stage later. In fact, this is oneof the absolutely most exciting part of the Topic Maps standard in myeyes, and sorely overlooked by the world at large. You read more about“locators” too, athttp://www.isotopicmaps.org/sam/sam-model/#sect-locators In fact, skim the standard itself; it’s quite full of good advice and
modeling ideas”–October 8, 2007 email from Alexander Johannesen
DECISION: All key identifiers are treated as properties of the entities (classes) they represent.