Jump to content

  • Log In with Google      Sign In   
  • Create Account

Interested in a FREE copy of HTML5 game maker Construct 2?

We'll be giving away three Personal Edition licences in next Tuesday's GDNet Direct email newsletter!

Sign up from the right-hand sidebar on our homepage and read Tuesday's newsletter for details!


We're also offering banner ads on our site from just $5! 1. Details HERE. 2. GDNet+ Subscriptions HERE. 3. Ad upload HERE.


Using Lisp (or another language) to generate fictional characters


Old topic!
Guest, the last post of this topic is over 60 days old and at this point you may not reply in this topic. If you wish to continue this conversation start a new topic.

  • You cannot reply to this topic
126 replies to this topic

#81 Neoshaman   Members   -  Reputation: 170

Like
Likes
Like

Posted 11 November 2003 - 06:54 PM

it seems that the 'program' structure can be done on paper before implementation (builduing tree struct) isn't it?? is there other kind of representation of the lisp before coding?
i mean visual one, if i understood a little what's going one with the doc i have read, i'm a visual people and work mostly before doing everything with graphic...
i just want to be sure (maybe some subtlety i miss in the syntax which fit better to a scriptural aproach than a visual one)

EDIT:
did bishop_pâss could move this to ai forum
i think it would fit better and wouldn't go 1O pages away in less than half a second
>>>>>>>>>>>>>>>
be good
be evil
but do it WELL
>>>>>>>>>>>>>>>

[edited by - neoshaman on November 12, 2003 1:57:32 AM]

Sponsor:

#82 kordova   Members   -  Reputation: 138

Like
Likes
Like

Posted 16 November 2003 - 09:19 AM

It''s sort of interesting, as I read Digital Mantras and experiment further with this system, I begin to see where the functions and system I''ve defined arenn''t simply limited to this domain. I can envision a twist on it being used in the system hinted at in the Plausible Interlude thread, as well as other things such as Natural Language and Knowledge Systems as the examples for the SHRDLU program in the book uses a system that, if not written in Lisp, certainly seems to stick to many of the same rules. Cheers.

#83 kordova   Members   -  Reputation: 138

Like
Likes
Like

Posted 16 November 2003 - 09:23 AM

quote:
Original post by Neoshaman
it seems that the ''program'' structure can be done on paper before implementation (builduing tree struct) isn''t it?? is there other kind of representation of the lisp before coding?
i mean visual one, if i understood a little what''s going one with the doc i have read, i''m a visual people and work mostly before doing everything with graphic...
i just want to be sure (maybe some subtlety i miss in the syntax which fit better to a scriptural aproach than a visual one)

I think you could certainly outline much of it on paper, though I guess it would sort of depend on your project. What do you mean by "is there other kind of representation of the lisp before coding?"? Also, which document are you referencing that you''ve read? In any event, cheers and your english is very good for someone so new to it.

quote:
Original post by Neoshaman
did bishop_pâss could move this to ai forum
i think it would fit better and wouldn''t go 1O pages away in less than half a second


I wish.

#84 bishop_pass   Members   -  Reputation: 100

Like
Likes
Like

Posted 16 November 2003 - 03:54 PM

quote:
Original post by kordova
I wish.
You think?



#85 kordova   Members   -  Reputation: 138

Like
Likes
Like

Posted 16 November 2003 - 04:13 PM

quote:
Original post by bishop_pass
You think?


?

#86 bishop_pass   Members   -  Reputation: 100

Like
Likes
Like

Posted 16 November 2003 - 04:17 PM

quote:
Original post by kordova
quote:
Original post by bishop_pass
You think?


?
I''m asking if you think it should be moved to the AI forum? How much better would it fare over there?



#87 kordova   Members   -  Reputation: 138

Like
Likes
Like

Posted 16 November 2003 - 04:20 PM

quote:
Original post by bishop_pass
I'm asking if you think it should be moved to the AI forum? How much better would it fare over there?


While, sadly, most people never leave the lounge, it'd probably do more justice to this thread to not be surrounded by and lost among random crap threads whereas it might occasionally be viewed in the AI forum. This thread's not incredibly active as it is obviously though I personally would like to contribute to it over time as I can.

I think this thread ties in nicely with the other associated threads and it would be neat to see them together, not that that will necessarily add to its popularity.

[edited by - kordova on November 16, 2003 11:23:07 PM]

#88 kordova   Members   -  Reputation: 138

Like
Likes
Like

Posted 17 November 2003 - 04:41 PM

Anyway, I have a few ideas and have some time to spend on the theory as well as implementing and refining the actual code so what do you think the next logical steps are? Are these functions/tools what you had in mind for this or are changes needed? Additions? I''ll probably try a few things I have in mind in the meantime. Cheers.

#89 bishop_pass   Members   -  Reputation: 100

Like
Likes
Like

Posted 17 November 2003 - 05:45 PM

Well, using similar data structures as already described, I would try and provide knowledge describing generic key events which happen in one''s life, with possible role fillers. Also, those key events would have slots such as ''can-trigger'' and ''often-caused-by'' which in turn point to other generic event types. We might have something like:

leaving-home:
subcategory-of: event
often-caused-by: parental-abuse wanderlust going-to-college
roles: leaver home

And then, we would have:

parental-abuse:
subcategory-of: condition
roles: child parent
abused: child
abuser: parent father mother

Then the system would examine these frames and create particular instances of them. For example:

leaving-home-john:
instance-of: leaving-home
leaver: john
home: johns-home

It''s rough, and just food for thought.

#90 kordova   Members   -  Reputation: 138

Like
Likes
Like

Posted 17 November 2003 - 05:57 PM

Alright, a lot of that is pretty much as I had expected. For the most part things exist as a sort of "tree" of "knowledge" as it were. Where I begin to lose myself in it, is once you''ve encoded so much information, how you manipulate it afterwards (at runtime I suppose). Again it''s most likely just my inexperience with more symbolic systems but I''m having difficulty in seeing a way to use this afterwards. Perhaps more thought will bridge the gap for me, as I do have seem to have something of a grasp on either end...

#91 solinear   Members   -  Reputation: 145

Like
Likes
Like

Posted 18 November 2003 - 06:42 AM

I look over this and being someone who likes to find complicated solutions, I think that maybe just LISP is the wrong path to start on.

I think that maybe a better solution might be to create a ''core set'' of characters, with attributes, traits and personalities, then build all other characters in relation to them, whether they be siblings or offspring. Also maybe create 10% ''external'' characters with little history related to the existing world that you''re using, to introduce some form of variety.

You''d have to highly define your character traits and attributes, strengths and weaknesses though and create some form of interaction system rules.

Example: Jose''s father is intelligent (156) and creative (130), but unfocused (direction of 79) and an alcoholic (critical weakness). Jose is not so intelligent (144), but he is structured (direction of 118) and realistic (make up a stat). You have to figure out how Jose''s intelligence is going to interact with his father''s and how his structure will react to the lack of focus and alcoholism. This will probably require some other form of AI (fuzzy logic, NN, etc...) that would then be fed into the LISP engine to create a more in depth history for Jose.

Just a thought from another angle.

#92 kordova   Members   -  Reputation: 138

Like
Likes
Like

Posted 18 November 2003 - 07:33 AM

quote:
Original post by solinear
I look over this and being someone who likes to find complicated solutions, I think that maybe just LISP is the wrong path to start on.

I think that maybe a better solution might be to create a 'core set' of characters, with attributes, traits and personalities, then build all other characters in relation to them, whether they be siblings or offspring.

You'd have to highly define your character traits and attributes, strengths and weaknesses though and create some form of interaction system rules.

You make a statement and give an example of your path but you didn't explain why that this would be superior. I think that system seems to shut in upon itself as soon as you start giving it those exact values from the beginning, but perhaps you will flesh out your ideas more to prove me wrong. (I believe this argument was made briefly earlier in this or a related thread...)

quote:
Original post by solinear
Just a thought from another angle.

Those are always welcome. Cheers.

[edited by - kordova on November 18, 2003 2:35:01 PM]

#93 bishop_pass   Members   -  Reputation: 100

Like
Likes
Like

Posted 18 November 2003 - 03:36 PM

quote:
Original post by solinear
Example: Jose''s father is intelligent (156) and creative (130), but unfocused (direction of 79) and an alcoholic (critical weakness). Jose is not so intelligent (144), but he is structured (direction of 118) and realistic (make up a stat).
I don''t get it. Only an RPG programmer would think like that, and for what purpose?

Storytellers don''t think like that. They think in terms of stories.



#94 kordova   Members   -  Reputation: 138

Like
Likes
Like

Posted 18 November 2003 - 04:29 PM

Did you catch the meaning of my poorly worded question, bishop?

#95 bjmumblingmiles   Members   -  Reputation: 122

Like
Likes
Like

Posted 18 November 2003 - 05:02 PM

Ok, I agree that the stat system goes against the point of your idea, but how does a computer keep Billy''s father straight? I mean, what if it spit out that he was an alcoholic and a MADD activist? Don''t the story elements need to some how be kept in sync? How can that be organized while avoiding personality factors of some type? By the way, what good LISP implementations are there for windows?

#96 Extrarius   Members   -  Reputation: 1412

Like
Likes
Like

Posted 18 November 2003 - 06:02 PM

You don''t have to use numbers to describe somebody. Just think of a person you know and describe them. You might use a number for height, weight, and age, but would you use their IQ, or rate how optimisting/pessimistic the person is on a scale some kind? You might say they are pessimistic, or very pessimistic, or a little bit pessimistic, but you don''t say ''pessimism level is 127''. A computer can be made to work the same way without crisp values.

Corman Lisp is a good lisp implementation for windows. You can get a free trial from http://www.cormanlisp.com/. The IDE is a 30 day trial(but you can still use it after that. You''re just not supposed to), but the actual compiler does not expire and is free for non-commercial development. I suggest you use Emacs for the IDE with the Corman Lisp compiler instead of the Corman IDE.

#97 bjmumblingmiles   Members   -  Reputation: 122

Like
Likes
Like

Posted 18 November 2003 - 06:10 PM

Yeah, but without a value, how can you describe varying degrees of something? I mean, Osama is evil, but McBride is very evil. What if there was someone in between? Without fuzzy logic of some kind how can that be stored? I guess since we are modelling human thought, it would have to be relative: McBride > Osama > Billy''s Father > Billy. But how would that translate to something useful?

#98 bishop_pass   Members   -  Reputation: 100

Like
Likes
Like

Posted 18 November 2003 - 06:18 PM

quote:
Original post by bjmumblingmiles
Yeah, but without a value, how can you describe varying degrees of something? I mean, Osama is evil, but McBride is very evil. What if there was someone in between? Without fuzzy logic of some kind how can that be stored? I guess since we are modelling human thought, it would have to be relative: McBride > Osama > Billy''s Father > Billy. But how would that translate to something useful?

Totally irrelevant. What''s relevant is what a person did. It''s not whether Osama is evil or not, or how pessimistic Joe Blow is, but what Osama did, or what Joe believes.


#99 RPGeezus   Members   -  Reputation: 216

Like
Likes
Like

Posted 19 November 2003 - 04:25 AM

You can''t confidantly call somebody evil based on the observation of their actions alone. This is the basis for the movie Unbreakable.

Mr. Glass looks evil at first, and appears to have done very evil things, but really he is just fullfilling his purpose in life, as evidenced by the existance of the Hero (Bruce Willis). Even if Bruce Willis had not existed, Mr. Glass would not have been evil. Conversely, Bruce Willis''s character is no more good than Mr. Glass is evil-- he just is what he is, through no will of his own.

So I guess what Im saying is that you can make someone good, or evil, but you will have to have your program arbitrarily decide these things and hope that the reader of your dossier agrees. Or, you can leave out Good and Evil altoghther, except for some obscure references to religion, and leave it up to the reader to decide.

Will




#100 kordova   Members   -  Reputation: 138

Like
Likes
Like

Posted 19 November 2003 - 07:13 AM

I don''t think bishop''s proposal called for such distinctions on behalf of the program.




Old topic!
Guest, the last post of this topic is over 60 days old and at this point you may not reply in this topic. If you wish to continue this conversation start a new topic.



PARTNERS