A recent email from Andrew S. Townley reminded me of a story I heard from my father decades ago.
Circa rural Louisiana, USA, early 1930’s. A friend had just completed a new house and asked the local plumber to come install the “commode.” When the plumber started gathering up his tool kit, the friend protested that he didn’t need to bring “all that” with him. That he had done this many times before. The plumber persisted on the grounds it was better to be prepared so he would not have to return for additional tools.
When they arrive at the new house, the plumber finds he is to install what is known to him as a “toilet.”
Repeating the term “commode” over and over again would not have helped, nor in a modern context, would having a universal URI for “commode.”
What would help, and what topic maps offer, is a representative for the subject that both “commode” and “toilet” name. A representative that contains properties that authors thought identify the subject it represents.
That enables either party to the conversation to do two very important things:
- Search for subjects in the way most familiar to them.
- Examining properties of the subject to see if it is the subject they were seeking.
One more important thing, if they are editing a topic map:
- Add additional properties that identify the subject in yet another way.
Understanding what others mean in my experience has been asking the other person to explain what they mean in different ways until I finally stumble upon one when I say: “I see what you mean!”
Topic maps are a way to bring “I see what you mean” to information systems.
*****
I am glossing over representatives containing properties of all sorts, not just those that identify a subject and that which properties identify a subject are declared.
What is critical to this post is that different people identify the same subjects differently and assign them different properties.
Intellectual credit for this post goes to Michel Biezunski. Michel and I had a conversation years ago where Michel was touting the phrase: “See What I Mean” or SWIM. I think my formulation fits the story better but you decide which phrasing works best for you.
Hi Patrick,
Thanks for the mention. Great post, and that’s exactly thi kind of thing I was talking about with my “dynamic identity” posts to the list recently.
I think your example explains it pretty well. One of the other important things about doing identity this way is keeping the traceability intact as well as isolating and eliminating those links on a temporary or permanent basis.
I think we’re in reasonable shape when adding this kind of information, but maybe not so good at taking it away. Always more to do, I guess!
Thanks again for the follow-up post.
Hope you have a great weekend! 🙂
Comment by Andrew Townley — November 13, 2010 @ 4:34 am
Andrew,
Thanks!
I think we are still weak at adding this sort of information for system subjects, like column headers or OWL sameAs, etc. All of those have equivalents in other systems as well. (To avoid my way or the highway approaches of integration projects.)
Good point about taking information away! Even if it were possible to see every news story on President Obama, most times I would not want to do so.
If I am a presidential policy adviser, I just want to see poll results for the last 24 to 36 hours. 😉 (humor for those of you who have trouble recognizing that sort of thing)
I will be posting on this soon but to some degree I think the current information “glut” is an artificial condition created by promiscuous aggregation that was then accepted as “that’s how it is.”
Your question remains a valid one even under reasonable aggregation conditions but perhaps more tractable.
Comment by Patrick Durusau — November 13, 2010 @ 8:39 am
I love this: “promiscuous aggregation”!!
That’s about the best description of the state of the art I’ve ever heard! 🙂
Look forward to your further thoughts.
Have a good weekend! 🙂
Comment by Andrew Townley — November 13, 2010 @ 4:38 pm