Mike said:
The way humans acquire language is precisely by starting not by reading Wikipedia but by mastering "fiction-like" sentences with simple subjects and simple actions and relationships - like "John sit" "John eat" "Jack like Jill. Me give Jill soap" etc. -based primarily in the here and now I can add a little to what Ben says in response to this issue. Although I am building Texai to be a student capable of being taught knowledge and skills, I am not starting with dog-like, nor baby-like, nor even with childlike behaviors. I want competency at a human adult level for a very limited set of allowed constructions; namely a controlled language designed solely as a bootstrap dialog system. Texai will thus acquire from a multitude of mentors the next increment of language knowledge and skills, and so on to get recursive self-improvement going. I spent the last six weeks rewriting my grammar engine for the third time in order to fix an unsoundness. Speed improvements have given me hope that a single modern multicore computer might even parse controlled language discourse into precise semantics faster than a human, i.e. faster than 5 words per second, and to be able to demonstrate its comprehension by answering questions about the subject text. -Steve Stephen L. Reed Artificial Intelligence Researcher http://texai.org/blog http://texai.org 3008 Oak Crest Ave. Austin, Texas, USA 78704 512.791.7860 ----- Original Message ---- From: Mike Tintner <[EMAIL PROTECTED]> To: [email protected] Sent: Monday, September 29, 2008 8:04:15 PM Subject: Re: [agi] universal logical form for natural language Stephen, One thing worth commenting on here is what seems to be your "non-developmental" concept of language acquisition. The way humans acquire language is precisely by starting not by reading Wikipedia but by mastering "fiction-like" sentences with simple subjects and simple actions and relationships - like "John sit" "John eat" "Jack like Jill. Me give Jill soap" etc. -based primarily in the here and now - and slowly building up stage by stage to ever more general and complex subjects and actions and relationships that transcend fictional/historical time and place - like "Human relationships are fraught with complications" and "Linguistics is focused on the study of sentences." Cognitive linguistics also lacks a true deveopmental model of language acquisition that goes beyond the first few years of life, and can embrace all those several - and, I'm quite sure, absolutely necessary - stages of mastering language and building a world picture. Stephen: Mike asked: How does Stephen or YKY or anyone else propose to "read between the lines"? And what are the basic "world models", "scripts", "frames" etc etc. that you think sufficient to apply in understanding any set of texts, even a relatively specialised set? Interesting that this question arises, given my recent poster presentation at the Fifth International Conference on Construction Grammar, and what I learned there. Accordingly, the following passage illustrates discourse and sentence constructions commonly found in fiction. One would not expect a Wikipedia article to have similar constructions. "John sat down in the carriage. His grim reflection stared at him through the window. A whistle blew. The train started shuddering into motion, and slowly gathered pace. He was putting Brighton behind him for good. And just then the conductor popped his head through the door." Given that the Texai bootstrap dialog system will have as its initial goal the acquisition of vocabulary mappings, grammar constructions, and dialog skills from human mentors, I am not currently giving much attention to solving Mike's example problem. >From what I learned at the conference, especially at the Prado-Alonso talk, >the author of a fictional passage often organizes the text so that the >focal object is introduced first "John sat ...". The discourse context >retains the focal concept from one sentence to the next, and also other >mentioned concepts subject to cognitive decay. Existing concepts are >positioned lexically to the left and new concepts are positioned to the >right. The word "carriage" needs to be disambiguated. In Wiktionary, >"carriage" has as its second word sense "a railroad car drawn by a >locomotive. Spreading activation from window to carriage should >subsequently rule out the first word sense for carriage "a wheeled vehicle, >generally drawn by horse power", assuming the KB entails the fact that a >railroad carriage has a window, and a typical contemporary horse-drawn >carriage does not. Processing "grim reflection stared" requires a solution for metonymy. I believe that discourse context elaboration and spreading activation can be helpful when determining the corresponding semantics for this form. To paraphrase, the system should figure out that "...grim reflection stared..." means the agent John perceived his own facial reflection and that the facial expression was grim. A more direct approach but less general is to simply teach the system a unique construction for each instance of metonymy, e.g. "...reflection stared..." <===> "there is a facial reflection which the agent perceives as staring" I could say more about the temporal ordering of the story sentences, but you all should get the idea about how Texai would read, and perhaps someday compose, fictional descriptive passages. -Steve Stephen L. Reed Artificial Intelligence Researcher http://texai.org/blog http://texai.org 3008 Oak Crest Ave. Austin, Texas, USA 78704 512.791.7860 ----- Original Message ---- From: Mike Tintner <[EMAIL PROTECTED]> To: [email protected] Sent: Monday, September 29, 2008 2:23:34 PM Subject: Re: [agi] universal logical form for natural language Ben and Stephen, AFAIK your focus - and the universal focus - in this debate on how and whether language can be symbolically/logically interpreted - is on *individual words and sentences.* A natural place to start. But you can't stop there - because the problems, I suggest, (hard as they already are), only seriously begin when you try to interpret *passages* - series of sentences from texts - and connect one sentence with another. Take: "John sat down in the carriage. His grim reflection stared at him through the window. A whistle blew. The train started shuddering into motion, and slowly gathered pace. He was putting Brighton behind him for good. And just then the conductor popped his head through the door." I imagine you can pose the interpretative questions yourself. How do you connect any one sentence with any other here? Where is the whistle blowing? Where is the train moving? Inside the carriage or outside? Is the carriage inside or outside or where in relation to the moving train? Was he putting Brighton *physically* behind him like a cushion? Did the conductor break his head? etc. etc. The point is - in reading passages, in order to connect up sentences, you have to do a massive amount of *reading between the lines* . In doing that, you have to reconstruct the world or parts of the world, being referred to, from your brain's own models of that world.. (To understand the above passage, for example, you employ a very complex model of train travel). And this will apply to all kinds of passages - to arguments as well as stories. (Try understanding Ben's argument below). How does Stephen or YKY or anyone else propose to "read between the lines"? And what are the basic "world models", "scripts", "frames" etc etc. that you think sufficient to apply in understanding any set of texts, even a relatively specialised set? (Has anyone seriously *tried* understanding passages?) ________________________________ agi | Archives | Modify Your Subscription ________________________________ agi | Archives | Modify Your Subscription ________________________________ agi | Archives | Modify Your Subscription ------------------------------------------- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244&id_secret=114414975-3c8e69 Powered by Listbox: http://www.listbox.com
