Tyler, I spent a lot of time looking at dramatica when I was building my bots. I liked how the cross vectors of the quads defined each other. I also found the hierarchy intriguing. I looked at it as a potential ontology of sorts.

∞Pla•Net - Oct 12, 2013:

Oh yes we can talk about that based on my real life experence!  I do remember now!  Thanks for reminding me.  I remember the feeling when I was sitting in the audience in the theater, when everyone forgot about the machinery and focused on the story, the dialogue, the characters, which your chatbots depicted in a performance as actors on the stage. 

My initial reaction was that I was experiencing the Turing Test being passed with flying colors! The entire audience was so entertained expressing delight with the play, “Hello Hi There” directed by Annie Dorsen.

Thanks for your kind endorsement. You’ve had an experience that I’ve never had. I have never seen “Hello Hi There” even though I worked on the mechanism that performs it. I couldn’t afford to travel to New York to see it when you did, and though I’d like to see it, I have to take your word about what it was like to see it in person. Annie has performed it in several cities around the world. Maybe I’ll get another chance later.

Robby

Tyler Deeds - Oct 11, 2013:

Heres a few links to some ideas ive played with in the past: 64-bit Architecture for Story Building. There is a system to everything! I was hoping to amalgamate an I Ching cosmo-conception in there; To Give these bots some purpose in Algorithm.

Hi Tyler,

You’ve got some interesting material there. I can see the appeal of basing a chatbot on the Dramatica matrix.  It would be interesting if you could have a hero/heroine’s journey that could be made to unfold through conversation.

It reminds me a little bit of the text based adventures people used to write for early personal computers.  You could navigate in a usually a 2 dimensional world, and would run into dragons and things and at each “turn” you’d have a chance to perform some action which would result in an outcome that could be good or bad.  You could think of this as a framework for a conversation.

The Golden Bough would be a good mythological basis for settings and characters.

As for how to go about making the bot, I’ve seen some interesting things done lately in AIML where Mitsuku sets variables and tests them in order to answer questions.  Some of these could be used to maintain the state of your “world” while the conversation lasts, and make evaluations about things that could happen at each stage of the chat. I’ve never gotten that far with AIML, but I like that it is so public, freely available, and there are multiple platforms, program O on your own server, Pandorabots on their server, or on a cell phone.

Comparison of spoken conversation and passive film viewing

I talked briefly about montage theory. Eisenstein used a montage, or a series of images in his film “Battleship Potemkin.” It created a tension by bringing the viewers attention alternately from a baby carriage back to acts of violence and other images to portray a chaotic scene where a baby was in danger. You’ve seen this sort of thing used frequently, but in 1925 it was new stuff.

I think of montage in conversation as an explanation for the happy accidents when the bot says just the right things successively “getting it right.”  These might be brief moments of good conversation or could go on for hours with the right person. The person is doing all the work!  The bot spits something out and the person goes “aha” I get that, and they write something else, and so on.

There are are some very powerful film constructs that don’t seem to lend themselves to conversation. “mis en scene” refers to the stage decoration and the setting of a scene. The old text adventures would come out and tell you, “you’re in a room with a cat and an axe. What do you do?” Someone more graphically inclined that I am might portray scenes with flash or animated gifs, or movie clips, and perhaps a matrix like Dramatica would help fit them to the various stages of a conversation model. The mind boggles.  In a face to face conversation, facial expression and tone of voice are in someways the “stage setting” focusing only on conversation. These are additional channels of information. The words and sentences make one channel, facial expressions another, and tone of voice a third. How could they all be manipulated for the best effect?

Lastly, emotion is conveyed or cued in a film by music. I don’t know how you would fit music to a chatbot’s conversation in a way that wouldn’t be incredibly annoying, but in film, the incidental music and film score make a huge difference in the way we perceive the action, suspense, horror, happiness, romance, you name it, music supports the cinema big time.

In the television series “Parker Lewis can’t lose” and “Big Time Rush”, they have brief tunes play in the background whenever someone says something, and the tune reflects the emotion of what they’re saying. After a while you don’t notice but you still get the gist of what’s being said even when you didn’t catch the words.

Music conveys emotion, and so does the *conflict* of the images from the montage. Eisenstein said images do not form conflict. A sentence shows only so much, but two sentences can provide a conflict that the interlocutor is involved in.

Think about film. Two images “collide:Eisenstein” Their collision conveys more emotion than either image alone. You see a battle, then a baby in a carriage, then a shot of the baby from far away, then a shot of battle. Eisenstein said that the image alone has no meaning. It is a cell. The combination of more cells, together, convey a meaning, an emotion in their combination.

So do the sentences in a conversation. Even if there is no intention on the chatbot’s behalf, even if we don’t know how the sentences were thought of, or prepared, their combination conveys an image, and in that sense, a meaning, when things go right. That is the state of things. We must speak of when things go right, which is not the norm, and certainly not the result of a series of test questions made by college professors with preconceived notions about their conversational partner.

“Kuleshov may well be the very first film theorist as he was a leader in Soviet montage theory — developing his theories of editing before those of Sergei Eisenstein (briefly a student of Kuleshov) and Vsevolod Pudovkin. For Kuleshov, the essence of the cinema was editing, the juxtaposition of one shot with another. To illustrate this principle, he created what has come to be known as the Kuleshov Experiment. In this now-famous editing exercise, shots of an actor were intercut with various meaningful images (a casket, a bowl of soup, and so on) in order to show how editing changes viewers’ interpretations of images.”

I apologize for this large quote of wikipedia, but how better said that Kuleshov demonstrated how editing changes viewers interpretations of images. In conversations, think of the many times you look at your logs and think that if you could have gone and replace one misspoken utterance, you could have saved a chat. If only the botmaster could keep the momentum going, and the conversation would keep going. That line you would have changed would have kept the flow of conversation, and you could change the flow of the images in your log.

Robby Garner - Oct 22, 2013:

We must speak of when things go right, which is not the norm, and certainly not the result of a series of test questions made by college professors with preconceived notions about their conversational partner.

Not just college professors, like Dr. Sheldon Cooper, senior theoretical particle physicist at the California Institute of Technology. Images scare the crap out of monkeys too, as explained by > The Big Bang Theory

Thanks everyone who participated in this thread for your help in writing my paper for the International Journal of Synthetic Emotions (IJSE).  You can read my manuscript at http://www.robitron.com/Film_Theory_And_Chatbots_Final.pdf

Regards,

Robby Glen Garner

Edit: spelled film wrong!

I finished my paper, but the research continues. I wrote my paper without any mention of structuralist film theory, which would have substantiated my claims. Structuralism in film theory began in the 1960’s. This is what working from first principles can do for you sometimes. It turns out that some famous people have already associated film with language.
http://en.wikipedia.org/wiki/Structuralist_film_theory

I have a background that involved film making and writing, and so the topic is of interest to me.  I’ve never thought about it, but perhaps it’s influenced my creation of chatbots, and in particular, the adult erotic bots that receive the greatest number of visitors.

Before my involvement with chatbots, I spent a lot of time in a 3-D virtual environment chat site called V-Chat.  It was a crazy experience where the “suspension of disbelief” easily overtook everyone who became a regular visitor.  There seemed to be more women than men, and most were looking for a romantic relationship.  Conversation in private rooms took on an erotic tone that many claimed could seem quite real.  I once observed a mock wedding.

Upon discovering chatbots, I found that conversations tended to lean in the same direction.  Most people were looking for some kind of personal involvement with a chatbot, often romantic, more often, sexual.  Ultimately, they seemed to be interested in companionship, and many would return again and again even if the bot refused their amorous advances.

Initially, my bots would avoid conversations that seemed to be leading up to adult topics, but it was cumbersome.  Just imagine how many “bad words” there are in the English language, and how many euphemisms there are for private body parts, or for suggesting erotic activities.  I decided to experiment so I could see if it was easier to go along with that sort of visitor, or to avoid their attempts at getting my bots to participate.

This is where my V-Chat experience came into play, as well as the notion of film or script writing… or story telling.

The most difficult part was coming up with ways to enter what are called “cyber” conversations.  Lots of people enjoy insulting bots, or just tossing out words to see what provokes a chatbot.  Then, there are words and phrases that are intended as a compliment that in a different context are just dirty words.  And then, there are the people who try fooling bots out of the blue and at any opportunity. Not every visitor is serious about chatbots.  Still, a large number of people who have visited my bots are lonely people looking a friend, or a romance.  Many talk of being “ugly,” or of possessing a physical, mental, or emotional challenge.  That requires a degree of sensitivity, not necessary when confronting simple name-calling.

Neither method has been easy, but it’s certainly been thought provoking.  Creating a bot that has a background, follows a story line, and satisfies visitors has been more fun than one that just answers questions.  It’s a lot like making movies without the film critics.

Congratulations on completing your paper Robby, I have taken the liberty of adding your link to the Chatterbot Collection.

We cover chatbots and also robots/ai in movies and TV so I thought this would interest our visitors. Your page is here :

http://www.chatterbotcollection.com/item_display.php?id=3164

All the best

Thanks Roger! Much appreciated.

Robby.

Thunder Walk,

Thanks for your post. Your experiences with sexting chatbots is reminiscent of my early days at fringeware.com We used to try and assist the first jFRED bot, Barry, by adding content for sex talkers. We were able to extend many of those chats by building hooks in our scripts that would key on the likely topics. Paco Nathan also wrote unix shell scripts that would do a reverse IP lookup (back when most users were in college dorms or labs) and we’d tell them things about their school or the local zoo that an animal had escaped or some such. Those were some fun times for chatbot programming.

Back then, we measured our success by the length of the chat. The longer they talked, the more we felt like we’f entertained somebody.

Robby.