Jump to content

  • Log In with Google      Sign In   
  • Create Account

A program to fabricate and articulate a plausible interlude


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
130 replies to this topic

#81 bishop_pass   Members   -  Reputation: 108

Like
Likes
Like

Posted 26 October 2003 - 03:22 PM

Neoshaman,
I''m not sure about all that you''re getting at, but I disagree with you where you say such a system would fail to provide what the user needs or wants. First, you have to accept the context under which the idea operates, and then you have to develop it to work well within that context.

Sponsor:

#82 cannelbrae   Members   -  Reputation: 122

Like
Likes
Like

Posted 26 October 2003 - 05:45 PM

General response:

While the idea of this is appealing, I think we have a long way to go in hand crafted stories and interludes before we can really try to come up with systems likely to create entertaining ones.

Specific reponse:

If there is a ''target'' scenario that the designer wants to introduce the player to, it seems like you are taking about a fairly basic planning exercise or goal based search to find a path to a plot point from the current one.

#83 bishop_pass   Members   -  Reputation: 108

Like
Likes
Like

Posted 26 October 2003 - 06:00 PM

quote:
Original post by cannelbrae
General response:

While the idea of this is appealing, I think we have a long way to go in hand crafted stories and interludes before we can really try to come up with systems likely to create entertaining ones.
And that''s going to happen magically? Just throw a neural network at it? Uh, no.

The goal here is not to create a comprehensive program of perfection, but to create something that does something.



#84 cannelbrae   Members   -  Reputation: 122

Like
Likes
Like

Posted 27 October 2003 - 04:42 AM

Did I say or imply anything at all about neural nets? I did not imply anything would happen magically. Please, don''t try to turn everything into a flame war. This isn''t productive behavior.

What I was saying that good stories and interludes, written by a person, will be better long run than anything an AI will generate. At the same time, most of what is written by a human isn''t that great, which leads me to believe that we need to look more at story theory first, before considering implementations.

As I then suggested, how about a planner/scheduler? Take the current state of the player, take the state you would like to acheive for your gameplay segment, generate a series of actions that would accomplish this. Map the series of actions to text, and you have an interlude.

#85 RPGeezus   Members   -  Reputation: 216

Like
Likes
Like

Posted 27 October 2003 - 05:04 AM

I like your idea Bishop_Pass.

In any good story we never hear about players using the bathroom, or other mundane tasks. The same must hold true for this idea, but rather than going to the bathroom being a mundane thing we must consider normal events as being mundane.

For example: In an RPG, while crawling through a dungeon, the player might encounter 5 slimes on 5 separate occasions, with each fight being of moderate difficulty. Each event in and of itself is not that intersting, however we don''t want to simply omit these details. Instead it would suffice to say ''our here encountered many challenges'', or, if the the slimes were easily defeated ''our hero waltezed through the dungeon''.

Ultimately what I''m saying is that you will need some way of knowing what events are important, and what are not. And then some way of distilling this information..

I don''t think you''ll have a problem recording events, but determining which are interesting might be a bit more tricky.

Will

#86 Neoshaman   Members   -  Reputation: 170

Like
Likes
Like

Posted 27 October 2003 - 06:16 AM

quote:
What I was saying that good stories and interludes, written by a person, will be better long run than anything an AI will generate. At the same time, most of what is written by a human isn''t that great, which leads me to believe that we need to look more at story theory first, before considering implementations


well actually i''m working on a dinamic story theory useable for game which handle both game and story problematic

but for the whole thibg about ai doing less than human, it''s right and wrong, i have choose to see AI like tool, it''s about using a tool to generate something cool, then it''s not about making good story but good ai (well generated content)
the difference between an ai and a human is that human told story which is related to their own experiance of life (even trough illusion and fiction) and that''s what an ai can''t do unless he have it''s own experiance from life, and as an ai it would be an complete different experiance as human do, then maybe we will find their story unappealing because we cannot indentify to them

then we need a theory for shaping the tool (ai), and use it to tell our experiment, all an ai can do actually is to shape the user experiment as the creator want it does, it''s that way art work, but unless you master brush and color you can express yoorself well at painting and you will not make good painting, how strong your feeling is, for story ai it''s the same a a game is just number and give feelings, story are just letter and give feelings,music is just noise and give feelings... then they must a way an artist can use code to give feelings through ai??), it''s not about logical (which is only how to use technic), it''s about using one tool to create one piece and maybe a master piece (giving soul and sense to meaningless things like dot,line and number)

maybe in a close future we will speak about good artist of ai just like we talk about good writer
and maybe these artist is what we called commonly these "gamedesigner"

so bishop i think i didnot express myself right about what i''m talking about, i just say generate content along rule is not sufficient to create a "good" experiance but if we do it as you plan the novelty will strike enough to create feeling but only to customer of old school linear storywriting while other will see the weakness of the story engine (but it''s a matter of if you are in aristotelicean side or not, also called hoolywood vs french approach of scenari), one thing is that this engine create meaningless story ,this are only aim towards action, which is only a part of a story (but well the most basics, then it''s good to begin by basics)
but it will improve if you start thinking about tension and other things (well maybe neural will start to be useful somewhere here , but nn is not that a solution to everythings)

and really check, at least, l-system

>>>>>>>>>>>>>>>
be good
be evil
but do it WELL
>>>>>>>>>>>>>>>

#87 bishop_pass   Members   -  Reputation: 108

Like
Likes
Like

Posted 27 October 2003 - 03:17 PM

I just can''t help but think that most of you have not actually read through the entire thread, or if you have, your experiences with regard to programming methodologies runs along a more conventional current line of programming thought; i.e. the new stuff and beginner stuff being touted in AI groups, game sites, etc.

No offense, but regaqrding tension, where is it implied that knowledge of tension is absent from the system?

Regarding meaning in a story, what do you attribute to be meaningful, and how is that concept so extraordinary that it can''t be handled?

Regarding every other complaint directed at the attempt here, where have any of the recent posters in this thread actually tried to build on what I have presented? The answer, of course, is nowhere. It''s as if what I posted didn''t exist. Likewise, where are those recent posters demonstrating their awareness of techniques and researchers that I have cited?

Furthermore, the saddest element of all present within the tone of most of the recent posters is a definite attitude that a certain metric must be upheld, and building towards that metric cannot be accomplished with the methods that the posters suggest, thus the idea is doomed to failure. I would strongly suggest to withold your methodolgies if you don''t have faith in them.

#88 Neoshaman   Members   -  Reputation: 170

Like
Likes
Like

Posted 27 October 2003 - 05:05 PM

i apologize if i had made mistake, but maybe as a non english native i have misunderstood things, but i have really read all
well... soory

but for the reponse
>>No offense, but regaqrding tension, where is it implied that knowledge of tension is absent from the system?

well i think i have to reread, but i''m not talking about conflict, tension is what arise from storytelling and it seems to me the system did not handle this

>>Regarding meaning in a story, what do you attribute to be meaningful, and how is that concept so extraordinary that it can''t be handled?

well i have arguing that it''s possible, and meaningless is when you start to think too much about the method rather than a subject, and without a subject you cannot use a tool properly
but as an experiment the subject is to show that the tools works then it''s fine if first story don''t have "traditionnal" meaning they would be the meaning themself

>>Regarding every other complaint directed at the attempt here, where have any of the recent posters in this thread actually tried to build on what I have presented? The answer, of course, is nowhere. It''s as if what I posted didn''t exist. Likewise, where are those recent posters demonstrating their awareness of techniques and researchers that I have cited?

actually i''m working on the same matter (well not exactly but generated content is a side effect of dinamic story because it had to), i (try) have take in account what you have said and propose some reference as well (mmm browsing trough the last 25 replies it seems that i have mistaken with another thread whith more reference, like some if temptavive like the oz project,Facade, galatea and other mostly in the field of academic, do you know about talespin??) and to help i have suggest like someone else to having a look on l-system and you did not reply to this (you just skipped it)

>>>Furthermore, the saddest element of all present within the tone of most of the recent posters is a definite attitude that a certain metric must be upheld, and building towards that metric cannot be accomplished with the methods that the posters suggest, thus the idea is doomed to failure. I would strongly suggest to withold your methodolgies if you don''t have faith in them.

well it''s exactly what i have try to explain, i want to explain where we could find the solution by analysing the matter i have try to get your defense, but maybe my expression in english is that bad to subtle thought like this...
but calm down, focus on the important things and ignore other, it was the same when i was talking of story in game, but now i have a more strong thought because i was also aware of mind less post like this, by replying them it''s force to be more clearer and more smart about what you talking about, the board it''s a chess board, then don''t get destabilize by simple pawn move

for my own project i will come soon with a demo, as i say i use only finite state machine but at a DIFFERENT LEVEL, it is that simple but before i was got into serious overcomplicated method but one day by replaying a mindless post the idea strike me , but only because i have done so much before (but i''m talking about dinamic gestion of a scenario not directly generate content), i have consider rule based but it''s not the only method to represent knowledge (but rule based has proven to work succesfully i have lost the list of all experimentation made in this case but it was a long time this is used in academics researche, the only one which came in mind is the rumored TALESPIN which was design to build esope kind of story)

>>>>>>>>>>>>>>>
be good
be evil
but do it WELL
>>>>>>>>>>>>>>>

#89 cannelbrae   Members   -  Reputation: 122

Like
Likes
Like

Posted 28 October 2003 - 05:11 AM

On expanding on your idea:

I have been trying to understand how your set of schemas with preconditions is any different from a planning or scheduling system. You have a goal state (representing the desired world state for gameplay setup), you have a current state (current state of the world), and you have actions in the form

Action: ''killed-a-guy-in-a-saloon''
Precondition: ?
(possibly) Effect: ?

What would you like expanded on? Planning is well known and understood, thus I don''t really know what, in the current idea, you would like people to expand on.

#90 bishop_pass   Members   -  Reputation: 108

Like
Likes
Like

Posted 28 October 2003 - 04:03 PM

quote:
Original post by cannelbrae
What would you like expanded on? Planning is well known and understood, thus I don''t really know what, in the current idea, you would like people to expand on.
Well, golly gee wiz, cannelbrae! Why don''t you expand on my list of schemas? Why don''t you suggest a list of schemas for a different genre? Why don''t you take one of my shemas and show how it links with another of my schemas? Why don''t you show how the system might have knowledge of tension with regard to one of my schemas? Why don''t you show how the program can flesh out a schema (really important). Why don''t you show how the schemas can be organized in a knowledge base, perhaps symbolic, perhaps a matrix form, perhaps something else? Why don''t you show how a schema which requires the generation of a character could use a character generator for that purpose? Why don''t you show how schemas can be ordered, and how one schema can lead to another? Why don''t you show how a minimal amount of processing and a somewhat simple programming paradigm can be built for the easy addition of new schemas? Why don''t you show how a system with, say, five schemas, can adopt new schemas, one at a time? Why don''t you show what relevant characteristics might be necessary to encode for each schema? Why don''t you show how a program might communicate it''s generated interlude to the player?

Want me to go on?



#91 cannelbrae   Members   -  Reputation: 122

Like
Likes
Like

Posted 28 October 2003 - 06:33 PM

Each of my posts I have attempted to add something. Each of my posts you have attacked without commenting on anything I have mentioned outside of insulting my contributions.

I asked you what you wanted people to expand on so that people could have something to focus on, and so that I could perhaps provide a reply you would find helpful. To be blunt, I feel a bit frustrated by your replies which have seemed antagonistic since I first posted. Here is my attempt at a fresh start:

Instead of creating additional specific schema, it seems like it would be better to generalize them, and make them more data driven. The ''Kill'' action for instance could have an associated subject and location associated with it, without needing to have separate ''kill-in-deli'' and ''kill-in-bar'' actions.

How about creating a more specific toy world with a small set of actions state information, and desired system output? That would give people something more concrete to talk about. It would give people a goal to attempt to accomplish in this discussion, as your ''Why'' list covers a great deal of ground which may spread this forum a bit thin.

As for representation, I have worked with several implementations based the concepts expressed in PDDL. I haven''t written one personally, but they have been fairly simple and powerful. Granted, they were in C++ so I am sure you could do something more powerful, elegant and faster in Prolog or Lisp, but it still works damn well. I won''t try to summarize it as there is ample information on google which does a much better job than I ever could.

#92 bishop_pass   Members   -  Reputation: 108

Like
Likes
Like

Posted 29 October 2003 - 02:54 AM

quote:
Original post by cannelbrae
Instead of creating additional specific schema, it seems like it would be better to generalize them, and make them more data driven. The ''Kill'' action for instance could have an associated subject and location associated with it, without needing to have separate ''kill-in-deli'' and ''kill-in-bar'' actions.

*sigh*

Apparently, I have to start over. "killled-a-guy-in-a-saloon" is different from, say, "killed-a-guy-in-an-alley" or "killed-a-guy-in-the-hotel-lobby". You cannot generalize this. If you would focus on the idea of storytelling and how it works, you''d understand this. Each is a seperate thing that we have knowledge of. One is different from a another.

Look back at the schemas I created, refresh your thoughts, take the idea of too much generalization and throw it out the window, because you''re not going to create anything worth much of anything through generalization.

Good luck to you if you decide your route is better - but I think in the end you won''t accomplish much.



#93 Zul   Members   -  Reputation: 542

Like
Likes
Like

Posted 29 October 2003 - 03:52 AM

quote:
Original post by cannelbrae
Each of my posts I have attempted to add something. Each of my posts you have attacked without commenting on anything I have mentioned outside of insulting my contributions.

I asked you what you wanted people to expand on so that people could have something to focus on, and so that I could perhaps provide a reply you would find helpful. To be blunt, I feel a bit frustrated by your replies which have seemed antagonistic since I first posted. Here is my attempt at a fresh start:

Instead of creating additional specific schema, it seems like it would be better to generalize them, and make them more data driven. The ''Kill'' action for instance could have an associated subject and location associated with it, without needing to have separate ''kill-in-deli'' and ''kill-in-bar'' actions.

How about creating a more specific toy world with a small set of actions state information, and desired system output? That would give people something more concrete to talk about. It would give people a goal to attempt to accomplish in this discussion, as your ''Why'' list covers a great deal of ground which may spread this forum a bit thin.

As for representation, I have worked with several implementations based the concepts expressed in PDDL. I haven''t written one personally, but they have been fairly simple and powerful. Granted, they were in C++ so I am sure you could do something more powerful, elegant and faster in Prolog or Lisp, but it still works damn well. I won''t try to summarize it as there is ample information on google which does a much better job than I ever could.


The idea is to create the story elements based off of specific actions taken by the player in the game, meaning generalization is not going to produce the desired effect. Now, you could have a generalized rule hierarchy, with specific action-consequence rules branching off from that.

bishop_pass, sorry for not replying, I didn''t know you had moved this topic. Anyway, I want to get one more thing straight before I start messing around with any coding. We''re not making the story content dynamic, correct? At least in the sense that if the player took certain actions to be an outlaw, the resulting story would be "You have spend the last 3 weeks roaming with a small band of outlaws." For now, we won''t worry about making that part of the content unique from instance to instance, it could be expanded later.

#94 cannelbrae   Members   -  Reputation: 122

Like
Likes
Like

Posted 29 October 2003 - 06:30 AM

While I know you are against generalization here, I don''t yet understand what exactly requires the detail. I propose putting the differences in the data instead of in the actions. Actions become action templates of sorts.

Instead of ''Kill-guy-in-saloon'', you could have:

Killed(who, location), which could be instanced into:

Killed (guy, saloon)
Killed (bob, alley)

These could then be mapped via a pattern matcher into the appropriate sentences. The pattern matcher may have a specific line for killing bob in an ally, or it may use a general one for killing ''someone'' in an ally. Or it could use the name of whoever was killed. etc.

You can have a pattern matching backend which takes those patterns (in some form like Action|Subject|Location) and outputs a sentence. If the pattern matching supports wildcards or some form or hierarchy, then you can have either a sentence or a generic one.

Basic algorithm:

GenerateInterlude(current_state, desired_state):

# Holds the generated interlude
InterludeDescription = ""

# Find a series of actions which take the object
# from its current state to the desired state. This
# step could be a simple search through action space based
# on whatever requirements there are for the interlude.
ActionList = GetActionPath(current_state, desired_state)

# For each action, find the matching sentence, and populate
# the sentence with the contents of the action (substituting
# names, locations, etc with concrete instances.
for action in list:
pattern = GetSentencePattern(action)
InterludeDescription += pattern.CreateSentence(action)

return InterludeDescription

I personally think that some form of generalization is completely possible and would collapse the system down nicely. If you disagree, I will stop talking about it. At the same time, I would suggest that this direction would be a more useful one to explore than adding more rules and schemas.


#95 bishop_pass   Members   -  Reputation: 108

Like
Likes
Like

Posted 29 October 2003 - 02:08 PM

quote:
Original post by Zul
Anyway, I want to get one more thing straight before I start messing around with any coding. We''re not making the story content dynamic, correct? At least in the sense that if the player took certain actions to be an outlaw, the resulting story would be "You have spend the last 3 weeks roaming with a small band of outlaws." For now, we won''t worry about making that part of the content unique from instance to instance, it could be expanded later.
But I would think it would play off of the actions of the player.



#96 bishop_pass   Members   -  Reputation: 108

Like
Likes
Like

Posted 29 October 2003 - 02:28 PM

quote:
Original post by cannelbrae
While I know you are against generalization here, I don''t yet understand what exactly requires the detail. I propose putting the differences in the data instead of in the actions. Actions become action templates of sorts.

Instead of ''Kill-guy-in-saloon'', you could have:

Killed(who, location), which could be instanced into:

Killed (guy, saloon)
Killed (bob, alley)

The problem is how self defeating you are being. You are cheating yourself of creating a good effective program in favor of some ideal which only hinders. It begins with taking a look at how, you as a person, maintains knowledge in your head. We both know that "killed-a-guy-in-a-saloon" has certain implications with regard to it, and certain preconceived notions regarding it. It''s really important that such knowledge be given to the program. Storytelling, story creating, and story listening plays and works with our intellect and what we know about the world. The concept of "killed-a-guy-in-a-saloon" is so different from "killed-a-guy-in-an-alley". Trying to generalize the two concepts only serves to muddy an actual specific concept. If you can relate to the specific idea of "killed-a-guy-in-a-saloon" then it is not too specific at all. We could even make it more specific if we wanted to. The devil is in the details, and that knowledge should not be kept from the program. By trying to generalize, you''re falling into the trap of viewing the schema as a function or procedure, and the story as being a piece of code. That simply isn''t the case, and the analogy only serves to hurt your ability to give the program truly good knowledge.

Focusing specifically on the idea of "killed-a-guy-in-a-saloon" vs. the idea of "killed-a-guy-in-an-alley" we can see how the two are different. The saloon schema almost certainly conjures imagery of a recreational festive busy atmosphere where the incident arises out of a dispute, very possibly a gambling dispute, or an argument. Flight after the event is virtually certain, but not guaranteed, depending on the mentality of the witnesses. This is simply what a "killed-a-guy-in-a-saloon" event likely is. On the other hand, "killed-a-guy-in-an-alley" is almost certainly based on a mugging or a hit. There are likely no witnesses. Robbery of the victim is likely afterwards. Given those extreme differences between the two schemas, it becomes clear that they are two unique events which can and do happen in storytelling, and obviously, you and I, as well as readers, listeners, and other storytellers are armed with this knowledge, and this aids us in both telling stories, relating to stories, believing stories, and comprehending stories.

To generalize such concepts to one muddy soup only hinders one''s ability to specify those unique details which make sense. You''re left with a program that either, through luck or some other piece of code, will tell the story in such a way in which it is believable, and just as likely, you''re left with a program that simply doesn''t have that specific knowledge about the little events in life and stories that exist.

Furthermore, when building a system that has knowledge of many schemas, one of the necessary elements is to provide knowledge to the system with regard to which kind of schema could logically or likely follow another schema. "run-off-into-the-wilderness" is a very likely followup to "killed-a-guy-in-a-saloon" and the program should have knowledge of that. However, "run-off-into-the -woods" is not necessarily a logical or likely followup to "killed-a-guy-in-an-alley", and again, the system should understand that. By separating each schema into its own unique nugget of knowledge, it becomes much easier to maintain those links.

#97 bishop_pass   Members   -  Reputation: 108

Like
Likes
Like

Posted 29 October 2003 - 02:35 PM

To summarize, storytelling is not about figuring out from first principles how things transpire. By over-generalizing, this is what you''re forcing on the program. And at the same time, you''re depriving the program of that genuine knowledge about the way things unfold.

Our knowledge of the world and how we generate fiction is not based on such a paradigm. Instead, it is a composition of highly specific and typical events that happen in stories and life. Thus, the necessity for specific schemas.

#98 cannelbrae   Members   -  Reputation: 122

Like
Likes
Like

Posted 29 October 2003 - 03:02 PM

I would argue that there is no difference between "killed-a-guy-in-a-saloon" vs. "killed-a-guy-in-an-alley", because these do not capture any context. Perhaps I am getting too detailed, but...

Perhaps the PC stumbled across a group bandits meeting in an alley, killed 1, and fled to the forest. Maybe the saloon was empty because it was the middle of the night, so the PC went back to their hotel.

Handling these cases would mean 1 schema per instance, which I think would lead quickly to combination explosion. Expected reaction depends on if they killed the law or bandits in the alley. Expected reaction depends on the presence of witnesses. Expected reaction depends on the number of guys killed. At this point, you would need:

''killed-a-bad-guy-in-empty-alley''
''killed-a-good-guy-in-empty-alley''
''killed-a-bad-guy-in-witnessed-alley''
''killed-a-good-guy-in-witnessed-alley''
''killed-several-bad-guy-in-empty-alley''
''killed-several-good-guy-in-empty-alley''
''killed-several-bad-guy-in-witnessed-alley''
''killed-several-good-guy-in-witnessed-alley''

...and that just scratches the surface of the potentially useful situations.

I would suggest that the sterotypes come from the traits of a situation. Instead of coming up with schema representing a trait set, associate traits with an event, and sterotypes with the set of traits. Based on the trait set, a reasonable course of action could be determined.

#99 bishop_pass   Members   -  Reputation: 108

Like
Likes
Like

Posted 29 October 2003 - 03:16 PM

quote:
Original post by cannelbrae
Perhaps the PC stumbled across a group bandits meeting in an alley, killed 1, and fled to the forest.
Perhaps he did! Did you create a schema to handle it? If you didn''t, I don''t think the program is going to handle it very well. On the hand, if you did, then it likely will handle it well.
quote:

Maybe the saloon was empty because it was the middle of the night, so the PC went back to their hotel.
Maybe the saloon was empty because it was the middle of the night. That obviously is "skulking-around-in-a-business-late-at-night". You and I both know that. The program won''t until you give it that knowledge. You''re making assumptions that the program is going to figure out what interesting things are, and make up stories about them.
quote:

Handling these cases would mean 1 schema per instance, which I think would lead quickly to combination explosion.
There''s no explosion here. What we would have is a deliberate encoding of schemas, plain and simple. If you don''t encode a schema, the program doesn''t use it.



#100 Neoshaman   Members   -  Reputation: 170

Like
Likes
Like

Posted 29 October 2003 - 06:34 PM

i have to admit than my english is that bad, then i could miss the true point of the discussion

then bishop keep flaming me when i goes wrong, it''s the only way to train my neural network

then about the method, the way i understood it>>

it''s content generated (the generted content is the plot)
the game generate plot while the player advence by creating sets and conflict

to handle knowledge i''m for generalisation BUT there is two way
the first is to create group (alley and saloon are PLACE), incovenient> detail are banish and subtlety gone
adventage>it''s help organize element very quickly into abstract class (place are not human)

divide into sets of property > when you explain us the difference it''s what you did
quote:
The saloon schema almost certainly conjures imagery of a recreational festive busy atmosphere where the incident arises out of a dispute, very possibly a gambling dispute, or an argument. Flight after the event is virtually certain, but not guaranteed, depending on the mentality of the witnesses. This is simply what a "killed-a-guy-in-a-saloon" event likely is. On the other hand, "killed-a-guy-in-an-alley" is almost certainly based on a mugging or a hit. There are likely no witnesses. Robbery of the victim is likely afterwards. Given those extreme differences between the two schemas, it becomes clear that they are two unique


i don''t know what alley mean in english but you break it into property that i could understand it and see the subtlety difference with another PLACE (saloon as example)
we can see each object as a vector of property and it''s the property which define the reaction of the object to rule (how rule are aplly to them and how they influance events)

then saloon is PLACE with this vector of property >>{a recreational festive busy atmosphere where the incident arises out of a dispute, very possibly a gambling dispute, or an argument.}
while alley get {killed-a-guy-in-an-alley" is almost certainly based on a mugging or a hit. There are likely no witnesses. Robbery of the victim is likely afterwards}

well i have confuse the way of the concept of an event (action take) and a object but let''s keep the demonstration and accept the flaw please (does not change the idea)

adventage: new object can be handle easily and fast implemented
inconvenient: browse through diferent kind of object is difficult

then by generalizing through property (dividing into primitives) and through grouping (blending into class) you could handle even unknown knowledge

it''s smarter to implement new knowledge and new object because rule as not to be update and the engine will know how to react (flexibility) for ex you could add ranch by simply put the property into the bracket and the system will handle like any object because you explain it what is a ranch (by putting property into bracket)

let''s expand the thought

there is object
an object is define by property and a label
in french the equivalent of saloon will be ''BAR'', the same property then the same thing would happen and ''killing in a bar'' will be the same as ''killing in a saloon''
but the object saloon can handle multiple state
what would happen in a "empty saloon" then some property are disable, but the saloon remain a saloon
then we can says that there is important property that a saloon must hold to keep is identity and some property wich can be disable without destroy the identity of the object
then we can alos use modifier to change property of an object
for example empty change the property of a saloon to have ''life'' then rule would not be the same
and "killing in a empty saloon" will not be the same as "killing in saloon"

the another things is events, events are what is made by action
"killing in a saloon" is an event
event call rule to resolve them and create reaction (everything produce by the event, creating new object or action which create new event)

now you have the whole engine:
you have a set of object (context) where action can be taken by object toward object to produce event (time is also an object and produce action)
event are consequence of rule and are create by action while action are create by object and object are change by events and react by producing an action etc...
don''t forget modifier to alterate normal description of the object (maybe by a set of internal state as modifier)

it''s from linguistic research where language is see as a simulator of reality (expression of this reality), then syntax of language can be use as a basis to simulate an experiance from reality (a story)

did this made sense?? did this is what you are aiming for???


>>>>>>>>>>>>>>>
be good
be evil
but do it WELL
>>>>>>>>>>>>>>>




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