Thursday, 2 July 2009

back on track...

Its been a strange month or so, trying to reassess where I am and whats happening with my phd.

In good news, my submission to the graduate symposium for the Creativity and Cognition Conference got accepted, along with some fairly substantial funding to go to California in October.

I'm rescaling what I am doing slightly, examining a particular domain as a test case for assessing creativity, then using this practical work to consider creativity more widely (although this consideration may not be practically realised in this PhD work).

One of my major problems with the assessment criteria suggested by Graeme Ritchie (2007, 2001) is that while it appears to be a methodology for assessing creativity, in practice it is too unspecified to produce meaningful results. As Ritchie is proposing a framework for anything creative, by definition he has to keep the framework generic and cannot define particular parts of the framework (such as: various measurements of the level of novelty in this program's output must be above the threshold of {theta} in order to satisfy certain criteria - so what is {theta}?)

My work suffered from the same criticism: by being very generalisable, it ended up saying almost nothing? It needed more surrounding context to ground it.

So now I'm concentrating on musical creativity again, specifically in jazz improvisation as its an area that interests me within music and there are a number of jazz generation systems of interest, which I am looking up:
  • Philip Johnson Laird's work
  • GenJam by Al Biles
  • Paul Hodgson's various jazz generation systems (described in his 2006 DPhil thesis)
  • Jeff Pressing
  • etc: I'm sure I'm missing lots out...

At the same time, I'm going to continue thinking about what is important for something to be considered creative.

Rather than looking for an over-arching and complete classification system for creativity though, I am looking for the important themes that are incorporated in creativity. With a few different practical projects on how to find these themes, added to my own intuitions on what is important, I'll identify a small set, test that set to see how it matches to human assessments of creativity, then re-juggle the set as necessary.

Graeme Ritchie's assessment methodology will be a starting point for me, although I think that what I end up with will be quite different from his suggestions.

I have to bear in mind that what turns out to be important for judging jazz generation systems may well be less important for other domains of creativity (in fact this is almost certainly going to be the case). But I shall cross that bridge when I come to it, rather than worrying about it too much in advance.

G. Ritchie. Assessing creativity. In Proceedings of AISB Symposium on AI and Creativity in Arts and Science, 2001.
G. Ritchie. Some empirical criteria for attributing creativity to a computer program. Minds and Machines, 17:67–99, 2007.

No comments:

Post a Comment