Silicate Wielder
January 4th, 2014, 12:33 AM
Hello again, im back with some more information on ATIS and my hopes for the project. I'll later be setting up a page for ATIS on my website that will include updates, news, and release information if it's requested (I hope you guys do, i'm working my ass off for this)
If you havent seen my first post on this please read that first: http://www.virtualteen.org/forums/showthread.php?t=199540
Before we get too into this I will also let you know that this post will be painfully long depending on how indepth I go into things, and how many times I decide to edit this.
To start off ATIS was something that originated back when I was in my repeat year of 6th grade and since then i've been going back and forth from it and other ideas i've had. by the time I was nearing the end of 7th grade I was doing early experimentation with online storage and retrieval of information, with things such as chatrooms, live feeds of information, sound networking, and a simple AI that learned and picked random responses as you spoke with it, After that was an attempt at creating a graphics engine that could print text out. And when I entered 8th grade I began to tinker with concepts for thinking systems and research on Linguistics, how we think, and toyed around with thoughts on how emotion could be simulated. I began researching and trying out AIs that spoke and responded. but never found one that could truly learn. Only recently have I begun true development, first with looking into past projects and isolating old code so that I could learn how to recreate it and make it more efficient, I then started creating the basic elements of the Response generator, starting with programing it to seperate text using a hardcoded divider, and then moving onto making it learn, and progressing to allowing it to learn multiple responses to an answer by directly asking.
And today marks the day when I enter phase 2 of response testing. I plan on it analyzing everything that is said to it, and allowing it to learn off of that.
However I do not quite know how I will achieve this, I just have some basic ideas of how processing will occour.
Right now what it does is takes user input, and then searches a long list of comparisons, if it finds one with that input, it then adds the resonse included to a list of responses to choose from, it repeats this untill it has no more to proccess, once done it is done, it either does two things, if there are no responses available it will directly ask for responses, otherwise it will choose a random response out of all of them and respond with that.
Tell me if there are corrections to be made with this post and tell me what you think, I need all the feedback I can get.
If you havent seen my first post on this please read that first: http://www.virtualteen.org/forums/showthread.php?t=199540
Before we get too into this I will also let you know that this post will be painfully long depending on how indepth I go into things, and how many times I decide to edit this.
To start off ATIS was something that originated back when I was in my repeat year of 6th grade and since then i've been going back and forth from it and other ideas i've had. by the time I was nearing the end of 7th grade I was doing early experimentation with online storage and retrieval of information, with things such as chatrooms, live feeds of information, sound networking, and a simple AI that learned and picked random responses as you spoke with it, After that was an attempt at creating a graphics engine that could print text out. And when I entered 8th grade I began to tinker with concepts for thinking systems and research on Linguistics, how we think, and toyed around with thoughts on how emotion could be simulated. I began researching and trying out AIs that spoke and responded. but never found one that could truly learn. Only recently have I begun true development, first with looking into past projects and isolating old code so that I could learn how to recreate it and make it more efficient, I then started creating the basic elements of the Response generator, starting with programing it to seperate text using a hardcoded divider, and then moving onto making it learn, and progressing to allowing it to learn multiple responses to an answer by directly asking.
And today marks the day when I enter phase 2 of response testing. I plan on it analyzing everything that is said to it, and allowing it to learn off of that.
However I do not quite know how I will achieve this, I just have some basic ideas of how processing will occour.
Right now what it does is takes user input, and then searches a long list of comparisons, if it finds one with that input, it then adds the resonse included to a list of responses to choose from, it repeats this untill it has no more to proccess, once done it is done, it either does two things, if there are no responses available it will directly ask for responses, otherwise it will choose a random response out of all of them and respond with that.
Tell me if there are corrections to be made with this post and tell me what you think, I need all the feedback I can get.