My research looks at how best to evaluate computational creativity systems - how creative is this computer program?
Computational creativity is when a piece of software acts in a way which would be perceived as creative if seen in a person.
From this research I have produced a sheet which can be used to evaluate and assess how creative a creative system is. It helps you judge the creativity of the computer program on several different aspects and components (see this previous post on identifying building blocks of creativity).
Here is the sheet.
Feel free to go ahead and use it, let me know how you get on. To cite this, please refer to my PhD thesis (title, more details and papers on this at my academic web site).
If you want even more detail, then you only have to wait till my doctoral thesis is ready in September 2011, or else please contact me (contact details on my web site) - I would love your feedback.
Thursday, 19 May 2011
Monday, 7 March 2011
Update: Building blocks of creativity
There has been a lot of useful feedback from my previous post reporting on my empirically-derived definition of creativity.
As a result, and after some more work that I have done, I've updated the results.
These factors form 'building blocks' of creativity - they contribute to the overall idea of what creativity is. This could also be thought of as an ontology of creativity - a collection of information relating to the nature of creativity.
To expand on these 14 factors, here is a diagram with a little more detail in each factor. (Click on the diagram to enlarge it)
Depending on what creative domain you are looking at (e.g. art, music, problem solving, proof generation, language use, design etc.) some of these factors will be more important than others.
From the results of a survey on musical improvisation, this is a breakdown of which factors are more important (and less so) in this type of creativity:
The next step in this work is to use the survey data and above results to evaluate and compare a number of music improvisation systems, to explore which are more creative than others and why. Results coming soon...
As a result, and after some more work that I have done, I've updated the results.
These factors form 'building blocks' of creativity - they contribute to the overall idea of what creativity is. This could also be thought of as an ontology of creativity - a collection of information relating to the nature of creativity.
 |
| Creativity is... |
To expand on these 14 factors, here is a diagram with a little more detail in each factor. (Click on the diagram to enlarge it)
Depending on what creative domain you are looking at (e.g. art, music, problem solving, proof generation, language use, design etc.) some of these factors will be more important than others.
From the results of a survey on musical improvisation, this is a breakdown of which factors are more important (and less so) in this type of creativity:
 |
| The importance of different components of creativity, in the context of creative musical improvisation. This illustrates the results of a survey of 34 people, spanning a variety of musical backgrounds from novice to professional expert. Note that some factors are considered to have some negative influence on creativity as well as a positive influence, e.g. Domain Competence. |
The next step in this work is to use the survey data and above results to evaluate and compare a number of music improvisation systems, to explore which are more creative than others and why. Results coming soon...
Tuesday, 8 February 2011
Today I defined creativity... Here's the definition
Here are the results of some work that I have been doing to capture in words a definition of creativity, as we understand and use the word 'creativity'.
[Ok so the timing in the title of this blog post is perhaps a little misleading... Rome wasn't built in a day and I didn't do this work all in one day... ]
The idea behind this work is that when we talk about what creativity is, certain words keep on cropping up as they are strongly associated with creativity. If we can capture those words, then they will collectively form a definition of creativity.
What I have done is to analyse discussions of creativity, to find what words appear significantly more often in such discussions (compared to discussions on unrelated topics). This gave me a list of 374 words. I've condensed this list by grouping words together that mean similar things and analysing the results, identifying 20 themes in the words.
Here are my results: 20 aspects of creativity. I propose that the combination of these 20 aspects collectively form a definition of creativity.
[Ok so the timing in the title of this blog post is perhaps a little misleading... Rome wasn't built in a day and I didn't do this work all in one day... ]
The idea behind this work is that when we talk about what creativity is, certain words keep on cropping up as they are strongly associated with creativity. If we can capture those words, then they will collectively form a definition of creativity.
What I have done is to analyse discussions of creativity, to find what words appear significantly more often in such discussions (compared to discussions on unrelated topics). This gave me a list of 374 words. I've condensed this list by grouping words together that mean similar things and analysing the results, identifying 20 themes in the words.
Here are my results: 20 aspects of creativity. I propose that the combination of these 20 aspects collectively form a definition of creativity.
What do you think?
 | |||||||
| MY DEFINITION OF CREATIVITY |
 |
| As an aside - for those who like to see the nuts and bolts of research: The main word clusters identified during this work, that assisted my work along the way |
Saturday, 2 October 2010
2nd International Conference on Computational Creativity (ICCC'11)
The call for papers is out for the 2nd international conference in computational creativity, to be held in Mexico in April 2011.
I'm planning to submit a paper containing my thesis work in a 'nutshell' (well, the work up to December 2010, when the submission deadline is).
Should be pretty good practice for summarising the key points of my thesis - plus if it gets accepted it will be a great way to promote what my thesis is going to be about, and get some feedback before submitting, with the people who are the target audience for my PhD. So far, people I've spoken to in this research community have been quite interested in what I'm doing, so hopefully there will be a good level of interest in my work at this conference.
I'm planning to submit a paper containing my thesis work in a 'nutshell' (well, the work up to December 2010, when the submission deadline is).
Should be pretty good practice for summarising the key points of my thesis - plus if it gets accepted it will be a great way to promote what my thesis is going to be about, and get some feedback before submitting, with the people who are the target audience for my PhD. So far, people I've spoken to in this research community have been quite interested in what I'm doing, so hopefully there will be a good level of interest in my work at this conference.
Monday, 26 July 2010
The first step on a long road, paved with LaTeX tiles and with a gleaming thesis at the end
Today I wrote the first words of my thesis.
Well...
For a while now I've had a thesis plan and a collection of chapter headings and subheadings - but no content. Now today I've started filling in the content.
What I've written today is very much for a first draft, full of [*** NOTES AND REMINDERS ***] and other aesthetically pleasing annotations. I should imagine that at least half of what I've written gets moved around, edited, or discarded, as I get more used to thesis writing.
But...
The first steps have now been taken on this PhD writing-up journey. After an apprehensive start of not knowing what on earth I was going to write as the first words, I just wrote something. Anything. Then changed it. Lo and behold, I'd started writing up. It feels good!
Well...
For a while now I've had a thesis plan and a collection of chapter headings and subheadings - but no content. Now today I've started filling in the content.
What I've written today is very much for a first draft, full of [*** NOTES AND REMINDERS ***] and other aesthetically pleasing annotations. I should imagine that at least half of what I've written gets moved around, edited, or discarded, as I get more used to thesis writing.
But...
The first steps have now been taken on this PhD writing-up journey. After an apprehensive start of not knowing what on earth I was going to write as the first words, I just wrote something. Anything. Then changed it. Lo and behold, I'd started writing up. It feels good!
Monday, 19 July 2010
expertise of programmer vs expertise of the programmer's program
It's not often I directly disagree with Maggie Boden but...
"only an expert in a given domain can write interesting programs modeling that domain"
[Margaret Boden 1994, What is Creativity? in Dimensions of Creativity p. 115]
(my brain is now busily plotting how to write a painting program)
A question of definition
Some questions have been going around my head recently, in the context of what creativity is. These questions have been along this theme: Are the defining characteristics of creativity actually just multiple recastings of the same thing?
- Can a discovery be useful but not interesting?
- Similarly, can a discovery be interesting but not useful?
This was inspired by Colton et al 2000, which looked at how 'interestingness' was evaluated by mathematical discovery systems. Here are some more developed thoughts: [Q. Can a discovery be useful without being interesting? I think NO in this domain because if some previously undiscovered concept or conjecture is useful then it has interest because it can be used. Q. How about in other domains? Depends on what interestingness means in those domains. Q. How domain-specific is interestingness-and how generalised can it be?In pure maths something is interesting if it helps you progress, therefore interestingness and utility are tied together this way. Q. In other domains, can discoveries be useful without being interesting? yes e.g. if they are a means to an end and if it is not your primary concern - I guess this applies to maths too - most maths conjectures are not interesting to me - unless I can see them being useful to me or in solving a notorious problem.Q. Can things can be interesting without being useful e.g. Doug Zongker's "Chicken" paper is interesting but not useful except as amusement (so does it have some value here in its humour - which is of course its main purpose? Hmmm I can't think of things which are interesting but not useful in some sort of way...)]
Continuing on this line of thought:- Is there any difference between things that are surprising and things that are novel [and can I just use novelty to explain both?]
[Surprisingness is linked into novelty - if something is seen before it is less surprising. But there is more to surprisingness than this - e.g. the result of a process may be surprising not because it is unseen but because it is derived in a different way - so... in a novel way...?]
ReferencesColton, S. and Bundy, A. and Walsh, T. (2000) On the notion of interestingness in automated mathematical discovery. International Journal of Human-Computer Studies (53) pp.351-375Zongker, D. (2006) Chicken Chicken Chicken: Chicken Chicken. Annals of Improbable Research (12) pp.16-21
This was inspired by Colton et al 2000, which looked at how 'interestingness' was evaluated by mathematical discovery systems. Here are some more developed thoughts:
[Q. Can a discovery be useful without being interesting? I think NO in this domain because if some previously undiscovered concept or conjecture is useful then it has interest because it can be used.
Q. How about in other domains? Depends on what interestingness means in those domains.
Q. How domain-specific is interestingness-and how generalised can it be?
In pure maths something is interesting if it helps you progress, therefore interestingness and utility are tied together this way.
Q. In other domains, can discoveries be useful without being interesting? yes e.g. if they are a means to an end and if it is not your primary concern - I guess this applies to maths too - most maths conjectures are not interesting to me - unless I can see them being useful to me or in solving a notorious problem.
Q. Can things can be interesting without being useful e.g. Doug Zongker's "Chicken" paper is interesting but not useful except as amusement (so does it have some value here in its humour - which is of course its main purpose? Hmmm I can't think of things which are interesting but not useful in some sort of way...)]
Continuing on this line of thought:
[Surprisingness is linked into novelty - if something is seen before it is less surprising. But there is more to surprisingness than this - e.g. the result of a process may be surprising not because it is unseen but because it is derived in a different way - so... in a novel way...?]
Colton, S. and Bundy, A. and Walsh, T. (2000) On the notion of interestingness in automated mathematical discovery. International Journal of Human-Computer Studies (53) pp.351-375
Zongker, D. (2006) Chicken Chicken Chicken: Chicken Chicken. Annals of Improbable Research (12) pp.16-21
Subscribe to:
Posts (Atom)
Posts
