Dale Pike has some interesting things to say about semantic focus as an organising principle for understanding technology – in particular for explaining how a specific aspect of some arbitrary technology helps with specific tasks. The down side of this, he observes, is that tools tend to become pigeon-holed by the application that is first used to explain them – seeing the tool in a different context might enable new uses but for many people there is a cognitive barrier set by the first mental model they have created.
He extends the thought to consider how context modifies the use we can make of specific pieces of information – as an example notes that are contributed to a topically-focused space such as a bulletin board or mailing list contrasted with the same note expressed in an individually-focused space such as a weblog. He sees syndication formats such as RSS as the connecting bridge that allows people to assemble published information into unique contextualised views that serve their specific needs.
This idea seems to be teasingly close to what I have described as projections of knowledge – each context is a map of the knowledge space projected in a particular way. Beyond the raw mechanics of content feeds the key to assembling projections/views is being able to find and select the information you want in an automatable way. The problem is to determine which concepts are “close” to each other on the map in question.
Most approaches that I have heard of use categorisation and filtering as a proxy for measuring conceptual proximity. Whether you use shared taxonomies or the more emergent “ folksonomy” approach a mechanism is needed to determine which labels are close to each other within the map of choice.
I can imagine this happening in a number of ways.
- At the most basic level tools could use some shared thesaurus to identify synonomous labels.
- An enhancement would be to allow the user to view a set of available labels and identify their own associations – this could in turn be published to allow “association aggregators” to form emergent thesauri.
- Even more subtle would be to allow the user to modify the view parameters by assigning votes to the returned concepts.
I have a hunch that all of this is buildable with currently-available standards. There may be tools out there already but I suspect they are proprietary – what we need are the simple building blocks to allow a “small pieces loosely joined” solution.