I have been talking about an AI program that will learn a simple natural language from text IO for some time. It would derive categories without using an a priori grammar. My claim is that because it is feasible to use a program that is able to *learn* the grammars of simple programming languages (without relying on an a priori grammar) then it would be feasible for the program to learn a simple more natural like grammar (without relying on a pre-programmed grammar). I have never been talking about using the grammars that are typically used with nlp for as long as I have known you, but I am interested in the computational methods that would allow a program to *learn* some grammar. I don't know how to explain this to you if after all these years you still do not understand what I am talking about. To put it another way, there are some common insights behind Pei's ideas and mine.
On Thu, Nov 14, 2013 at 5:48 PM, Piaget Modeler <[email protected]>wrote: > Also Jim, > > I've seen a lot of text but no architecture diagrams from you. Figure 1 > in the following > > http://www.aaai.org/ocs/index.php/SSS/SSS13/paper/view/5772/5944 > > might be a good architecture for what you're trying to accomplish. > > Roland Hausser's language proplets look very similar to Schilling and > Narayanan's X-Schemas. > Hausser also has Context schemas that don't include the surface lexeme > attribute. I think it's > well worth it to take a serious look at this paper and Hausser's two books. > > As for Pei's approach, I think it's quite interesting. I ran it by Roland > and he was intrigued as well. > > Cheers, > > ~PM > > ------------------------------ > From: [email protected] > To: [email protected] > Subject: RE: [agi] Re: Text-Based AI Should Be Able to Handle -Simple- > Context-Sensitive Language > Date: Thu, 14 Nov 2013 14:09:26 -0800 > > > Watch Pei's video at least and see. > > ~PM > > ------------------------------ > Date: Thu, 14 Nov 2013 15:09:53 -0500 > Subject: Re: [agi] Re: Text-Based AI Should Be Able to Handle -Simple- > Context-Sensitive Language > From: [email protected] > To: [email protected] > > > Pei Wang's video (wherein it's demonstrated that you don't even need a > grammar at all). Intriguing... > > https://www.youtube.com/watch?v=XAFt3o6x-KU&list= > PLZlLHCryX93J5O2iGzkSd7HjRKU9kb0tF&index=31 > > > I don't understand what you are getting at - as it relates to what I have > been talking about. > Jim > > > > On Thu, Nov 14, 2013 at 1:42 PM, Piaget Modeler <[email protected] > > wrote: > > I would recommend a couple of Roland Hausser's books: > > > http://www.amazon.com/Computational-Model-Natural-Language-Communication/dp/354035476X > > > http://www.amazon.com/Computational-Linguistics-Talking-Robots-Processing/dp/3642224318/ > > and > > Pei Wang's video (wherein it's demonstrated that you don't even need a > grammar at all). Intriguing... > > > https://www.youtube.com/watch?v=XAFt3o6x-KU&list=PLZlLHCryX93J5O2iGzkSd7HjRKU9kb0tF&index=31 > > > Thoughts? > > ------------------------------ > > > > Date: Thu, 14 Nov 2013 12:39:00 -0500 > > Subject: [agi] Re: Text-Based AI Should Be Able to Handle -Simple- > Context-Sensitive Language > > From: [email protected] > > To: [email protected] > > > > > Context Free and Context Sensitive grammars are based on > > http://en.wikipedia.org/wiki/Chomsky_hierarchy > > A programming language is based on a strong foundation of a > > context-free grammar. Imagine a text-based AI program that would be > > able to learn through trial and error. I believe that it is possible > > to create such a program that would be able to learn a very simple > > programming language - like simple database commands. Then, relying on > > an incremental argument, I am saying that it should be feasible to > > write a similar program that could learn a simple context sensitive > > language. But the references to the formal grammars are only meant to > > help you understand what I am trying to get at. As I read the > > Wikipedia entries I realized that my use of the technical terms was > > not quite right but I feel that it is ok because I was really talking > > about a simple natural language. If it is feasible to write an AI > > program that can learn a simple programming language then it should be > > feasible to write an AI program that could learn a simple version of a > > 'natural' language by using the simpler database commands. Why isn't a > > database able to learn a simple natural language? Because the ability > > to learn is a prerequisite. > > I also made a reference to type IV language in the thread, but I > > should have said a type 0 language (or type 0 grammar). In fact my > > argument is based on the fact that a program which was able to learn > > some simple context-free database commands would be able to use those > > commands to learn some simple context-sensitive grammars. So I am > > really speaking of a Type 0 language. > > > > > > On Fri, Oct 25, 2013 at 10:17 PM, Jim Bromer <[email protected]> > wrote: > > > I am convinced that it would be easy to get a text-based Learning-AI > > > program learn to respond in fairly simple ways to simple texts. (And > > > I will be in a position to try it out in the near future.) The > > > question is whether this kind of ability has to be at the expense of > > > an ability to integrate more sophisticated kinds of learning into it. > > > > > > I just do not see why people have not produced solid examples of > > > simple learning using text-based AI unless the problem was either that > > > they felt they needed to impress the skeptics or they became > > > confounded by their own, more complicated use of language. > > > > > > Simple language does not have to be at the level of a programming > > > language. I think that programming languages are "context free" > > > because even though the apparent context may seem to violate the > > > context of the substrings taken separately, any particular string > > > (that is any grammatical string) will still only generate one > > > particular output. > > > > > > So a computer could (genuinely) learn about simple strings that might > > > not be context free and use them to generate different points. As > > > long as this was kept relatively simple it should be completely > > > feasible and it might be a good starting point to examine what was > > > going on. (Even though a text only AI program would not be capable of > > > applying its knowledge in a sophisticated way, it could still > > > constitute genuine learning in my opinion because it would be able to > > > learn new things within the domain of the text-based interactions.) > > > > > > So even though my data management system is neither simple nor > > > sophisticated, I believe that I will be able to use it for simple but > > > somewhat sophisticated kind of learning which would be general within > > > the limits of the domain of text. > > > > > > Jim Bromer > > > > > > ------------------------------------------- > > AGI > > Archives: https://www.listbox.com/member/archive/303/=now > > RSS Feed: > https://www.listbox.com/member/archive/rss/303/19999924-4a978ccc > > > Modify Your Subscription: https://www.listbox.com/member/?&; > > Powered by Listbox: http://www.listbox.com > *AGI* | Archives <https://www.listbox.com/member/archive/303/=now> > <https://www.listbox.com/member/archive/rss/303/24379807-f5817f28> | > Modify <https://www.listbox.com/member/?&;> Your Subscription > <http://www.listbox.com> > > > > > -- > Jim Bromer > On Thu, Nov 14, 2013 at 1:42 PM, Piaget Modeler <[email protected] > > wrote: > *AGI* | Archives <https://www.listbox.com/member/archive/303/=now> > <https://www.listbox.com/member/archive/rss/303/19999924-4a978ccc> | > Modify <https://www.listbox.com/member/?&;> Your Subscription > <http://www.listbox.com> > *AGI* | Archives <https://www.listbox.com/member/archive/303/=now> > <https://www.listbox.com/member/archive/rss/303/19999924-4a978ccc> | > Modify <https://www.listbox.com/member/?&;> Your Subscription > <http://www.listbox.com> > *AGI* | Archives <https://www.listbox.com/member/archive/303/=now> > <https://www.listbox.com/member/archive/rss/303/24379807-f5817f28> | > Modify<https://www.listbox.com/member/?&;>Your Subscription > <http://www.listbox.com> > -- Jim Bromer ------------------------------------------- AGI Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-f452e424 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-58d57657 Powered by Listbox: http://www.listbox.com
