thanks for the reply’s. 

So It’s sound’s like the MODbot would be the brain/filter of the other bots and not adding to the conversation itself. I have no real idea, I assumed the bots would be able act as their own “Filter”. So would each bot need a MODbot of there own? It just occurred to me, not trying to be funny, but whats going to keep the MODbots on point?

I was hoping to load character parameters according to each characters profile goals and have that influence the dialog. If she a good mom with kids, in times of trouble, she will be putting her kids first. If shes a drunkard she will be putting alcohol first. If I cant get the bot dialog trick to work I’ll probably use a controlled variation of variables and just like I’m going to do with the bots just keep hitting the remix/re-talk button, keeping what I like and remixing the rest till I get something I like.

All the dialog from the stories (and actions for that matter) will have many descriptive tags for categorization including emotional descriptive tags of the four basic emotions happy, sad, afraid/surprised, and angry/disgusted these will have levels 1-5 if they have any emotions at all. That way it will be easier to correlate the stories with what they are supposed to be feeling in the story, but that’s a long way off.

the idea that I’ve proposed might look something like this:

(Please refer to the image located at http://geekcavecreations.com/images/CBO/StoryBotDiagram.jpg )

In this diagram, there are 3 “bots”, though only two of them have direct access to output. the third “bot” is merely a moderator/guide that takes its inputs from the outputs of the other two, and from a “timeline script”, which is more or less a collection of criteria and conditions required to be met before the story can advance. The moderator (let’s call it ModBot)looks at not only this list of conditions to be met, but also the outputs of the other two bots, and feeds both the output of each bot (StoryBot1 and StoryBot2), as well as commands or instructions, to the appropriate bot. For example, if StoryBot1 says something that furthers the storyline or meets a condition in the timeline script, the ModBot will alter the list of conditions for that aspect of the timeline and issue both SB1’s output and any relevant instructions to SB2, which would then process that data and formulate an output of it’s own. The ModBot would then analyse that output to see if any conditions were met, and if not, issue new instructions to guide SB2 to cause it to either alter or expand upon it’s output; OR it could possibly feed SB2’s output, along with possible other instructions, back to SB1. this part of the plan is still a bit “fuzzy” to me.

As to the “timeline script”, as I mentioned, it’s not a “second by second” list of events, but more a collection of conditions and criteria, organized into linear “chunks” that cannot be accessed out of order, and cannot advance until all conditions are met. The types of conditions and criteria could range from specific events, to the “utterance” of a certain phrase, or even weather conditions, or time of day. This may require the use of other modules or bots that generate random conditions, or perhaps a module that tracks “time of day” within the story. The possibilities are virtually endless here.

Ok, getting ahead of myself here, sorry. In addition to the link, above, I’m attaching the image in hopes that it will show up within the post. In my previewing of the post it wasn’t visible, so the link is further insurance.

Yes I see… very interesting Thanks I’m definitely going bookmark this thread and refer to it when the time comes

You have my email, Steve, so if there’s anything I can do to help, just holler.

Something came to me after I posted that, so I thought I would share. It might be interesting to consider adding a third chatbot to the mix to act as narrator, so that it’s not just the two “voices” weaving the story. By having a narrator, the story becomes less of a conversation and more of an actual story, which would (IMHO) make for a more engaging experience. Sure, it would make for a lot more work, but I think it would be well worth the effort.

I agree a third “bot” steering things from the background is a good idea.

Steve M - Nov 16, 2016:

I was hoping to load character parameters according to each characters profile goals and have that influence the dialog. If she a good mom with kids, in times of trouble, she will be putting her kids first. If shes a drunkard she will be putting alcohol first.

Short of developing strong artificial intelligence, I don’t know that there is a way of doing this. There haven’t been successful story generators made yet, most experiments just try to align sentences with other sentences that contain related words. None of that research uses character goals and decisions or anything so sophisticated. That would take a detailed understanding of events, recognising problems, solving problems, planning… all still underdeveloped areas in A.I., and quite some leaps further from a simple story generator.

I’m sure you can have characters with a list of priorities (food, shelter, kids, alcohol), but I wouldn’t know how to link that to an appropriate choice of phrases. Your third bot should probably be the one to generate the storyline problems and choices for the characters to pick from. Like a choose-your-own-adventure book maybe.

This is not an official reply but I just wanted to comment that It won’t be a story generator. it will be a tool, an Intelligent Short Story Instrument. I’ll need to be right there making decisions as I run it. I’ll try to automate it as much as possible but this program will not think, it just has to do what it can, waiting for me advise it.

Some of the Machine learning will be watching how I advise it. Looking for patterns, and with my approval making more rules according to the patterns it sees

Getting some really good input from you guys and I’ll definitely take it. I’m going to make a mock up this weekend so it’s better explainable. This is the hardest part getting it out of my head turned into something feasible, however impossible it sounds at first.

This might be of interest:

Annie Dorsen’s “Hello Hi There”

http://anniedorsen.com/showproject.php?id=6

This project uses a neural net to generate text, that can vaguely, remotely, somewhat, be mistaken for stories:
https://www.literai.com/tutorial