Wednesday, February 16, 2005

More folksonomy

I've just been keeping my eyes on the folksonomy pros and cons discussions (there's a nice overview article on Burningbird) - obviously I'm in complete agreement with regard to flat social categorisation replacing classification systems, but think that it is all a question of value judgement: the folksonomy approach combines personal information retrieval (my tags) with social information retrieval (whereas formal classification will not account for a personal mental space), it connects what you know directly with what others know without you having to give up your worldview to do so. Secondly it allows for the rapid evaluation of a knowledge space - this may not be optimal classification, but it made me think of John Nash and the Nash Equilibrium (as seen in A Beautiful Mind): I know this is not the equilibrium per se: (so apologies for quoting scripture to the Devil's purpose) but the best strategy for a group as a whole may be a compromise solution that is not the optimal solution from a choice. Watch the bar scene in the film with the blonde, and you'll see what I am getting at. With a folksonomy, what we lose in precision, we may gain in rapidity and flexibility as well as connectivity - it's a simple trade-off, and therefore its applicability will depend upon circumstances.
So I've been pondering a little bit more as to how to move from the loose spontaneity of folksonomy to a more controlled vocabulary, and this Porter stemming tool from hackdiary presents one possibility, I suppose. Stemming could be run against a folksonomy (rather than just against a personal user's tagset) to give an initial suggestion as to how to consolidate the tags within - for instance by suggesting synonymous terms. How this information is then dealt with is another matter (consolidating the tags would have a negative impact on personal tagsets, whereas a controlled vocabulary would allow the personal tags to stay, but also be mapped as synonyms etc. to other, "standard" tag descriptors, is my assumption)...

No comments: