I've been re-discovering today this first exchange with Jack and others, about five years old now, amazed and quite pleased to find out we seem to keep following the same track, and to agree basically with most of what I wrote at the time. Don't know if that is supposed to be good news or bad news, though...
Actually there was an interesting notion introduced then that we have unfortunately a little forgotten since, which is subjects as attractors in conversations. This we should consider again, along with other mathematical tools linked to quantic superposition we have been discussing lately. On the same lines, I had an exchange a few days ago with Michel Biezunski who is currently exploring the field of fiber bundles as a possible tool for subject representation, in the line of his recent presentation at Extreme Markup 2005.
You comments about subject identity and attractors reminded me of a discussion I had a few months back with Steve Newcomb and Bryan Thompson. In particular:
ReplyDeleteI see Google Suggest as providing a kind of attractor basin around a key family subspace. As you type in a keyword that makes sense in your own frame of reference, Google Suggest, "suggests" other topics that align with the source key family or property set. While Google is working with unstructured words, I think a similar approach would be very helpful in funneling a user into a more structured context for identifying data, or bridging different domains.
The ideas are not really new, and I've seen things like this done with Topic Maps, but the new implementations are slick, and as people get comfortable with these systems, there are more opportunities for extending them.
This is in response to a message that I have posted on my long term memory (blog) at Identity Semantics in XBRL and SDMX