Academic Research

I’m Getting My PhD in…

I wanted to write a little about my research. You can read a bit about it here. The diagram above is what the early system looked like. I’ve moved beyond that in a couple aspects, mostly because I’ve gone back to the drawing board a couple times.

So, what am I trying to do? Well, it’s hard to explain, but I suppose I’m trying to decompose storytelling and then logically reconstruct it. I’m basically building a storyteller or perhaps a story-re-teller. In Computer Science, these are called “story generators.” Some of the earliest work done in this area is by Vladimir Propp. If you want to play around with some form of Propp’s theories, here it is. If you want to read up on stories and “Expressive Processing,” I suggest you read Noah’s book.

I, myself, should take a second look at the story systems described in “Expressive Processing.” Off of the top of my head, however, the most popularly referenced story generators are Universe, Tale-Spin, and Minstrel. Besides the Proppian generator you can find online, the only other story generator I’ve ever used was my labmate, James Skorupski’s generator, Wide Ruled. I would say that difference between the Propp generator and James’s Wide Ruled could be the difference of transportation for a tricycle versus a motorcycle.

Now, I’ve read about a number of story generators: Mark Riedl’s IPOCL and Rafael Perez y Perez’s Mexica. Those two specifically designate themselves as story generators. Now, there are also a number of additional systems that focus on the interactive experience, such as: Peter Wehrauchs’s Declarative Optimization-Based Drama Manager, Mei Si’s Thespian, and James Lester’s U-Director. There are also rhetoric driven systems such as: Abelson’s Goldwater Machine and Michael Mateas’s Terminal Time. All of these projects focus on narrative and character, where generators focus on the story, drama managers focus on the interactive experience, and rhetoric-emphasized systems focus on how stories are told.

Honestly, there is some lineage of where all this work comes from, but it’s still a loosely defined area. Having just discovered Mendeley, I’m going to put much work into refactoring my understanding of the literature, but now, you have some understanding of what my work is based on.

For my own sake, here’s a list of future posts I want to write…

  • Faith healing from the Foundation of Digital Games Conference to Bethel Celebrities to international houses of prayer
  • What the heck is prophecy?
  • Churchianity vs Charismania
  • I Got the Gift of Tongues at an Artificial Intelligence Conference
  • This talk: http://vimeo.com/12725793
  • Decline of the world
  • Beautifully Insignificant
  • Revival 1010
  • Stanford Jam Session
  • When “Jesus Wept”

9 thoughts on “I’m Getting My PhD in…

  1. The video you posted is really interesting– not sure I have time to watch it all right now, but hey, epistemology is always fun. Some of his points seem very good, others maybe a bit naive, but that may just be b/c he’s trying to translate philosophical problems into everyday language. Also, I like the reference to Propp– have all sorts of things to say about structuralism (both syntagmatic and paradigmatic) and fairytales or other narratives. Will have to get back to you later bout that…

    Hopefully these links will work– you might find these books interesting:

    On epistemology: http://www.goodreads.com/book/show/838513 (the author is both a scientist and a philosopher, which gives him a neat perspective– a bit old now, but I like what I’ve read so far)

    On Fairy Tales (this is 800 pages long, but I can point you towards the interesting bits if you find a copy): http://www.goodreads.com/book/show/706596.Interpretation_of_fairy_tales (this is the best book on fairy tales, period– not just structure, but it does a great job bringing analysis of structure and analysis of meaning together)

    And this is a bit unrelated, but I’m interested in what you would think of this book: http://www.goodreads.com/book/show/606199.On_the_Internet

  2. @Carl,

    Regarding the Stackhouse video: Yea, I really liked the broadness of his talk, and I believe it uses a neat thought experiment to approach a question that everyone’s asked before.

    Regarding Propp: A well know and respected researcher in this area once said “Everytime someone mentions propp.. I want to vomit.” http://twitter.com/ffpaladin/status/17982141345
    The problem being that Propp is way over-used.

    In regards to stories: I’m actually using the Bible to deconstruction/reconstruct stories. It seems almost too ridiculous to be true. All I can say is, BRING IT!

  3. Oh, and I think I need to gripe about the Proppian tale generator (which I’ve seen before but never really looked into)– the text it fills in is as un-fairytale-ish (OK, not a real word) as can be, which I guess just underlines the fact that more is involved than a series of functions (though I admit Propp’s formalist approach is pretty well suited towards generating material, even if it started as description).

    Other factors that could help (maybe)– the essential “moves” (collections of functions) and their semantic value (what Holbek talks about– basically combining syntagmatic or surface structure with paradigmatic or deep structure), plus typical motifs and tale types (there is an index for each), and a better understanding of fairy tale stylistics (Axel Olrik wrote an article on the “Epic Laws of Oral Narrative” 100 or so years ago– old, but a decent intro). Holbek covers style, and the generation of symbols, pretty well, along with the structuralist stuff. The problem, of course, is that the goals of description and analysis are very different from those of generating a text.

    I think the problem/perspective of this sort of project is very interesting– we might compare it more to a course in fiction writing, rather than a literature course, but I don’t think the analogy works perfectly when it comes to creating programs which generate stories– it might be more like this: if a literature course is the drawing of a map based on an encounter with a real plot of land, then creating a story algorithm is like generating a real landscape from a map. I think we could also compare this to Searle’s Chinese Room problem: http://plato.stanford.edu/entries/chinese-room/. I confess, I’m not so familiar with Searle, and don’t even have time to read the link I just posted– but I think we can see a link to Reception Theory, where meaning is generated on the side of the audience (http://www.goodreads.com/book/show/4049340-reception-theory). In a way, it’s comparable to the way writing itself works– the author “means” when he/she writes, but once on the page, the text is nothing but ink and paper– not even “text” really. It’s on reading that it “means” again. This gets complicated when you try creating an algorithm which generates a story– but in the end, the goal isn’t to create a true “story teller”, ie, an intelligent entity with expert competence in story telling, but rather to create something which consistently puts out material which a competent reader can “read” as a coherent story– which doesn’t make the task easier, but maybe reframes it a bit.

    Waugh, sorry, that became really long, and I’m sure you’ve thought about most of these problems. Just the first chance I’ve had to really think through the connections between this work and what I do…

  4. @Carl,

    It’s funny, I just wrote a chapter for the book,”Halo and Philosophy,” titled: “Halo, AI, and Believable Experiences: Would Cortana Pass the Turing Test?”

    In that paper, I address the Chinese Room argument, and I basically find it to be moot. I’ll post the chapter sometime soon.

    I’m building something for function, so I’m less concerned with how it does something. I’m more concerned with whether it does what I want it to do.

    Here’s the paper on that: http://portal.acm.org/citation.cfm?id=1822309.1822326

    Thanks for the link on Reception theory. It feels quite relevant to what I’m doing.

    ——

    BTW, I’m calling my new system,”GENESES,” b/c I’m deconstructing stories from Genesis.

  5. Great title. 🙂 I’ll try to get to the paper– I think I saw it before but wasn’t sure I’d be able to follow it (too different from my usual perspective, and with a whole different lingo), but I’ll give it a go. Later though– right now I have to figure out what to do about a student who cut and pasted the first paragraph of his essay on Beowulf from Sparknotes…

  6. Curveship is not a story generator. It is a work in Interactive Fiction. However Nick and myself published a paper together joining our systems. You can find it in the proceedings of the IJWCC 08.
    Pablo Gervás and Federico Peinado worked in a story generator based on Propp.

    I think you should check my paper where I compare Mexica, Brutus and Minstrel. It might be useful to you.
    I think it is important to clearly distinguish those papers that model the creative process, from those where the designers of the system pre-define the possible outcomes.
    For me this is a core difference.

    Are you sending a paper to the ICCC-11?

    Good luck in your research.

  7. Hi Rafael!!

    I’ve looked at your work!

    You know, I took a quick look at the Curveship site and saw the work “Generating,” and just categorized incorrectly. The problem I kind of run into is that there are a bunch of story systems that are all quite similar, with varying foci.

    Yes, I’m looking to submit something to ICCC, maybe I will see you in Mexico!

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s