GTVH.. additional thoughts

Of late I've been battling with the age old question of authorial intention in relation to the semantic-script theory (or SSTH) of Raskin and the more developed general theory (or GTVH) of Attardo and Raskin.

I've been thinking about my supervisor's work on the structures of political speeches and, in particular, the speeches of Churchill. The semantic and linguistic structures look so much like the structures of humour. Couplets and triplets working, often using incongruous elements, to make a memorable and surprising statement. Maybe this is why they can be so easily parodied? This leads me to question whether the application of the General Theory of Verbal Humour presupposes the intention of the script or performance. Would the GTVH give a 'false positive' if applied to "Now this is not the end. It is not even the beginning of the end. But it is, perhaps, the end of the beginning."

Anyway, these are the ongoing thoughts.

Bots I have known...

This post is retrospective account of the technological work I've done in first three weeks of 2010.

The burning question remains... How do I get two bots to talk to each other?

My early attempts were with Program N (AIMLpad - http://216.128.99.72/AIMLpad/). This, after some considerable fiddling allowed me to see two bots talking. I had to learn some Perl scripting and get the memory allocations sorted out. However, in the end, it almost worked. Why only almost? It seems that if there is too much recursion in the AIML sets the script fails to generate a response. So, after a variable number of interchanges, the interaction would fall to blank lines.

I could have kept going down this path but I doubt I have the programming skills necessary to seriously sort the problems.

Also, I noted in the documentation that what the Perl script was doing was providing a means of imitating what a web server could do.

For that reason I decided to bit the bullet an have a serious look at Program D. I've run version 4.1.5 quite successfully - it uses a built in 'Jetty' mini-web server. Cool. Version 4.6 of Program D has had the web server removed to make it 'easier' to implement on a real server.

To get version 4.6 working on my Mac was a very steep learning curve. Eventually I fixed my local file association errors and got the simple GUI (Graphic User Interface) console working. There was much cheering! Then I started on the .war file - this is a web application that contains the 'engine' and the 'brain' of the bot. My logic was - If I can get one of these to work, then I can get two, then I can use Javascript or Actionscript (?) to get the two to talk to each other.

The process went something like this:
1. get Apache Tomcat server working on my machine.
2. get ANT ( a compiler program operating).
3. Recompile the .war file from my working version of Program D 4.6.
4. Deploy on Tomcat.

All good? Well I've got the .war to deploy but it's throwing an error message I can't yet understand.

I know so little about programming but I feel the need to know. The tools of the trade have an effect on the product - so I need to better know my tools.

More soon.

So far

Linguists approaches.. some notes

These are not fully formed thoughts... After doing some reading it has occurred to me that the Attardo / Raskin approach is predicated on a couple of not incontestable assumptions.

1. The Freudian distinct between "innocent'' and "tendentious'' jokes. I tend to have a Derida-esque dislike for binary oppositions. The General Theory of Verbal Humor (GTVH)tends to focus on tendentious humour by including a category for the butt of the joke or script.

2. The approach seems to assume that humour is easily and consistently recognised. We all know what humour is, right? What happens if GTVH is applied to a non-humorous text? Does it fail to find humorous structures? Or, does it only work on those text which are socially or culturally deemed to be humorous?

Just thoughts at this stage.

The more I read...

In The Consolations of Philosophy, Alain de Botton, talking about metaphors said Seneca believed that, "Arguments like eels: however logical, may slip from the mind's weak grasp unless fixed there by imagery and style. We need metaphors to derive a sense of what cannot be seen or touched, or else we will forget" (p.92).

Humour is a slippery and mercurial topic of study. The range of theoretical approaches is daunting. Disciplines as diverse as linguists, sociology, anthropology, psychology (in all of its many guises), communication, literary studies, and, even mathematics have contributed to the field. All of them, to greater or lesser degrees, lay claim to being at the centre of this funny universe.

What has occurred to me of late is that the study of humour may well be liken to the 'Wave Particle Duality' problem of physics. Light, and it seems all kinds of matter, can be viewed as being composed of particles or waves.

"As experiments were performed and evidence accumulated, the implications quickly became clear and alarming:

Light functions as both a particle and a wave, depending on how the experiment is conducted and when observations are made.

The most common interpretation is that the wave function represents the probability of finding a given particle at a given point... Particles end up distributed according to the probability laws, and therefore exhibit the wave properties. In other words, the probability of a particle being in any location is a wave, but the actual physical appearance of that particle isn't".

http://physics.about.com/od/lightoptics/a/waveparticle.htm (accessed 13/01/2010)

At times humour can be studied in the the most 'particular' manner - the linguistic approach of Raskin, Attardo, et al. This provides a very specific view of the inner working of a script. However, at other times, humour is studied as an analog for (and of) other elements. Humour is a wave form that sweeps across our psychic, social and cultural landscape.

The only constant seems to be that we all recognise that humour exists and we have the ability to differentiate the humorous from the humourless.

I'll leave the last words to WC Fields - "The funniest thing about comedy is that you never know why people laugh. I know what makes them laugh but trying to get your hands on the why of it is like trying to pick an eel out of a tub of water". (There are those eels again.)

http://thinkexist.com/quotation/the_funniest_thing_about_comedy_is_that_you_never/331459.html (accessed 13/01/2010)

The Literature Review Plan

The reading is slowly progressing. My search for an accessible text that introduces Raskin and Attardo linguistic theory has yielded surprising results. Victor Raskin has edited a Humor Theory Primer 2008. I've only read part of the introduction but it seems to be a 672 page literary review! It covers the historical development of humour research from multiple disciplines as well as dealing, in some detail, with the material I was looking for. The version I've found is through Ebooks Library - for as little as $176 I can get a PDF version of the beast. This seems like a good start for Christmas reading.

To be honest, the size of the literature review seems to be growing with every page I read. This seems to be the normal course of events.
However, the over-all plan is roughly as follows:

1. Humour Theory - an historical overview incorporating contributions from psychology, philosophy, sociology, anthropology, linguistics, communication, literary criticism, mathematics etc. Philosophy needs to get a guernsey due to Bergson's contribution to the field - even if it isn't the most accepted theory in the current debates.

2. Artificial Intelligence - again an historical overview that traces the development of natural language agents. Largely this would deal with Reeves and Nass Media Equation, its detractors; the Turing Test, and its detractors. The aim is to situate my work outside Artificial Intelligence Computer Science domain and place it in a media context.

3. Scriptwriting - I'm undecided about this section covering heuristic comedy writing texts - these usually have some very loose basis in a theoretical perspective but they tend not to have much credibility in academic circles. Also, I'm hard pressed to think of an example that isn't inextricably tied to a single media form, e.g. writing for TV or radio or film etc.

4. Creativity research - This I would also approach from an historical perspective coming to a conclusion that a confluence model of creativity that incorporates the individual, social and cultural elements is most appropriate for academic research. Would this section be better used as an overarching theoretical perspective, particularly if I can make a successful hybrid with Actor Network Theory (a colleague has recently pointed me at a good resource for this)?

Is all of this possible in 10000 - 12000 words and in two months? I'm hoping that January will be largely clear for writing. Realistically I don't think I can get all of tis done. But the first section is the one I would like to attack first - I feel this is my weakest area and I need to get some handle on it.

ECA Forefront Conference - 2009

Two weeks ago I delivered a presentation to the Early Career Academic conference at the University of Newcastle. The presentation was largely an updated version of my PhD confirmation presentation. The updated sections included an expanded overview of theories of humour and a discussion of how to implement a pair of chat bots.

The feedback was quite positive. However, the feedback that surprised me was that my audience (mostly people from a Science and IT background, with some Communication folk) were interested in the idea of empirical testing. How will I know that the bot interchanges are funny? How can this be tested? Should I use a test audience as part of my research? Should I ask 'experts' (members of the 'field' to use Csikszentmihalyi's term) like stand-up comedians and writers to give their opinion?

This is not something that I had initially planned. And I'm in two minds about it. At this stage I would prefer not to go down this path. Rather, I plan to use an established technique for understanding the structure of humorous exchanges, something like the linguistic / discourse techniques of Raskin and Salvatore (SSTH - Script Theory and/or GTVH - General Theory of Verbal Humor).

This will develop with more reading in the area.

Which way to handle the brains

Part of the work I've been doing over the past couple of week has been devoted to planning how to get the bots to talk to each other.

Option 1: This involves using two web servers, one for each character. Each would have a version of the Alicebot 'engine' (ProgramD version or Charliebot) and have its own AIML set. The timing of the interchanges would be controlled through a Flash interface.

Option 2: This calls for only one instance of the Alicebot 'engine' and uses the Flash interface to swap from one AIML set to the other. Both Dick and Detail Boy share the 'engine' but the outputs would be predicated on which 'mind' is being used.
At the moment, this sounds the most practical.





Option 3: I had a brief infatuation with the idea of creating a version of the 'engine' entirely in Flash in Actionscript 3.0. This would have allowed the greatest flexibility but would have soon become a programing and management nightmare. A colleague talked me out of it - thanks to Dr Keith Nesbitt.




In later posts I'm going to explain why I've chosen to base my work on the Alicebot 'engine' and AIML rather than other chatbot structures.