From - Wed Aug 29 09:48:49 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: WELCOME TO THE CSE 676 NEWSGROUP Date: 28 Aug 2001 18:22:31 GMT Organization: University at Buffalo CSE Department Lines: 32 Distribution: sunyab Message-ID: <9mgnh7$427$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:18 ------------------------------------------------------------------------- Subject: WELCOME TO THE CSE 676 NEWSGROUP ------------------------------------------------------------------------- Welcome to the CSE 676 Newsgroup. Please feel free to post questions, comments, announcements of relevance to the course here. I will do the same. I may also, as appropriate, repost private emails that you might send me if I deem them to be of general relevance to the rest of the class, but I will always hide the identity of the individual who sent me the email. ------------------------------------------------------------------------- William J. Rapaport Associate Professor of Computer Science & Adjunct Professor of Philosophy Member, Center for Cognitive Science Associate Director, SNePS Research Group (SNeRG) 226 Bell Hall (office: 214 Bell) | work: 716-645-3180 x 112 Dept. of Computer Science & Engineering | fax: 716-645-3464 SUNY Buffalo | home: 716-636-8625 Buffalo, NY 14260-2000 | rapaport@cse.buffalo.edu ------------------------------------------------------------------------- CSE: http://www.cse.buffalo.edu/ homepage: /~rapaport/ SNeRG: /sneps/ Center for Cognitive Science: /pub/WWW/cogsci/ Buffalo Restaurant Guide: /pub/WWw/restaurant.guide/ From - Wed Aug 29 09:49:14 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: SYLLABUS UPDATE Date: 28 Aug 2001 18:27:33 GMT Organization: University at Buffalo CSE Department Lines: 9 Distribution: sunyab Message-ID: <9mgnql$45e$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:19 ------------------------------------------------------------------------- Subject: SYLLABUS UPDATE ------------------------------------------------------------------------- I have slightly updated the syllabus to include a reference to relevant sections of Russell & Norvig's AI text: http://www.cse.buffalo.edu/~rapaport/676/F01/syl.html#russell From - Wed Aug 29 09:49:34 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: TEXTS Date: 29 Aug 2001 13:23:41 GMT Organization: University at Buffalo CSE Department Lines: 36 Distribution: sunyab Message-ID: <9miqct$9th$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:20 ------------------------------------------------------------------------- Subject: TEXTS ------------------------------------------------------------------------- Rob Milewski writes: > In reference to the CSE676 Syllabus - "On Reserve at SEL" > > This book can be ordered through amazon.com via zshops > Readings in Nonmonotonic Reasoning > by Matthew Ginsberg > http://s1.amazon.com/exec/varzea/ts/exchange-glance/Y01Y0306049Y6957097/qid=9990 > 39864/sr=1-1/102-7106253-7326557 > > This book can be ordered through amazon.com: > Knowledge Representation (Special Issues of Artificial Intelligence, an > International Journal) > by Ronald J. Brachman, Hector J. Levesque, Raymond Reiter (Editor) > http://www.amazon.com/exec/obidos/ASIN/0262521687/qid=999041457/sr=8-1/ref=aps_sr_b_1_1/102-7106253-7326557 > > Robert Milewski 510-6091 > Research Scientist > Center of Excellence for Document Analysis and Recognition > State University of New York at Buffalo ========================================================================= Thanks Rob! A couple of comments: 1. Besides zshops, you might try alibris.com 2. The 2nd book you mention is NOT one of our texts. It was a special issue of the journal _Artificial Intelligence_ devoted to KR, but only 1 article in it is of direct relevance to us: the one by Hirst, which I've put on reserve. (This volume is NOT the out-of-print B&L anthology.) From - Tue Sep 4 09:38:08 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: CVA & MedNLP Projects Date: 31 Aug 2001 19:04:15 GMT Organization: University at Buffalo CSE Department Lines: 20 Distribution: sunyab Message-ID: <9mon3f$mha$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:21 I have reorganized the CSE 676 website to make it easier to find information relevant to the CVA and MedNLP projects. In addition, I have added some new info on CVA that Karen Wieland, from the Grad School of Ed, just sent me. This includes some websites that you can search for texts that will have passages in them containing "target" words, and -- of great importance -- lists of target words that students of certain reading levels probably wouldn't know. So: in addition to the passages I discussed in lecture today, an alternative is to choose one of the words on these lists, then find passages, either on your own or via the other websites mentioned above, that contain that word. Take a look at: http://www.cse.buffalo.edu/~rapaport/676/F01/cva.html http://www.cse.buffalo.edu/~rapaport/676/F01/mednlp.html From - Tue Sep 4 09:38:27 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: TERM PROJECT QUESTION Date: 4 Sep 2001 12:55:10 GMT Organization: University at Buffalo CSE Department Lines: 16 Distribution: sunyab Message-ID: <9n2ive$htj$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:22 ------------------------------------------------------------------------- Subject: TERM PROJECT QUESTION ------------------------------------------------------------------------- A student writes: > In order to complete the project for the course 676, > is it necessary to write a computer program to model it? > If not, what should we have to do? It is not necessary, but if you choose to represent your text in SNePS, it would be preferable to program it in SNePS. Later, I will assign the SNePS tutorial, so that you will all be able to do this. If you choose not to write a program, then you should show your representation in some other way. From - Wed Sep 5 09:48:15 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: NEW STUDENTS Date: 4 Sep 2001 17:09:48 GMT Organization: University at Buffalo CSE Department Lines: 27 Distribution: sunyab Message-ID: <9n31ss$pub$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:24 ------------------------------------------------------------------------- Subject: NEW STUDENTS ------------------------------------------------------------------------- If you are new to CSE 676, please do the following: 1. Fill out the "name sheet" at http://www.cse.buffalo.edu/~rapaport/676/F01/nameSheet.txt and turn it in at the next lecture 2. Be sure to read the syllabus, which will be constantly updated, at http://www.cse.buffalo.edu/~rapaport/676/F01/syl.html 3. Be sure to monitor the newsgroup at news:sunyab.cse.676 or at least to monitor the newsgroup archive at: http://www.cse.buffalo.edu/~rapaport/676/F01/news.txt From - Fri Sep 7 09:13:11 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: QUERY ABOUT C.V.A. WORDS Date: 7 Sep 2001 12:55:12 GMT Organization: University at Buffalo CSE Department Lines: 14 Distribution: sunyab Message-ID: <9nag3g$6va$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:25 ------------------------------------------------------------------------- Subject: QUERY ABOUT C.V.A. WORDS ------------------------------------------------------------------------- A student writes: > Upon looking at the two websites that you posted for 676, I > noticed some words that are way above the age group we are trying to > target. Some words are for 16th grade. :) My understanding is that anyone younger than 16th grade (i.e., senior year in college) would be unlikely to know those words. But note that we are interested in all age groups, not just middle schoolers. In fact, the original grant proposal called for studying 2 groups: middle school and college. From - Fri Sep 7 09:13:17 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: UPDATED SYLLABUS Date: 7 Sep 2001 13:12:22 GMT Organization: University at Buffalo CSE Department Lines: 8 Distribution: sunyab Message-ID: <9nah3m$7c4$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:26 ------------------------------------------------------------------------- Subject: UPDATED SYLLABUS ------------------------------------------------------------------------- A typo in the Tentative Schedule on the syllabus has been corrected; see: http://www.cse.buffalo.edu/~rapaport/676/F01/syl.html#dates From - Tue Sep 11 09:47:22 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: USING PROLOG (OR JAVA, OR ...) AS A KR LANGUAGE Date: 10 Sep 2001 14:05:06 GMT Organization: University at Buffalo CSE Department Lines: 32 Distribution: sunyab Message-ID: <9nihai$964$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:27 ------------------------------------------------------------------------- Subject: USING PROLOG (OR JAVA, OR ...) AS A KR LANGUAGE ------------------------------------------------------------------------- A student writes: | Is Prolog a viable choice for a representation | language? I know that for the most part it follows the rules of | first-order logic, but with some extra constraints. Many of you have asked similar questions about your favorite programming language. The question is a bit like "Is Prolog (or whatever) a good way to represent lists?" There's a category mistake here: Lists are a data structure that can be represented in any *programming* language you want. To do that, you need to use the basic (or primitive) data types of the language to implement lists. That's easy in Lisp or Prolog; harder in, say, Pascal. SNePS is not a programming language; it is a KR language. It happens to be implemented in Lisp, but that's more or less a historical accident. We are studying KR languages. So, you need to choose a KR language; you can then implement that (if you wish) in your favorite programming language, perhaps in order to make use of built-in reasoning abilities of the programming language. Prolog has such built-in reasoning abilities; Lisp, Java, etc., do not. SNePS is a KR*&R& language, so it has its own built-in reasoning abilities. If you choose to represent your passage "in Prolog", what you would really be doing is representing it in a Horn-clause restriction of first-order logic (that's right: you would *not* have the full power of FOL), and then using Prolog as an inference engine. From - Tue Sep 11 12:35:54 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: FURTHER COMMENTS ON CHOICE OF A KR LANG Date: 10 Sep 2001 14:32:32 GMT Organization: University at Buffalo CSE Department Lines: 16 Distribution: sunyab Message-ID: <9niiu0$a3f$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:28 ------------------------------------------------------------------------- Subject: FURTHER COMMENTS ON CHOICE OF A KR LANG ------------------------------------------------------------------------- Rather than deciding *now* that you'd like to program in Prolog, or Java, or ..., you should concentrate on analyzing your passage in the way we did in class on Friday. Later, you can decide what KR language to use. Your choice should depend on, among other things, which KR language has the appropriate representation facilities for your analysis of your passage. For instance, if you decide you really must represent things in a modal, second-order language with defaults, Prolog (or Java or ...) almost certainly won't allow you to do this (at least, not without a *lot* of tweaking). (SNePS, however, will ;-) From - Thu Sep 13 12:37:30 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676,sunyab.cse.575 Subject: OFFICE HOURS Date: 12 Sep 2001 15:05:02 GMT Organization: University at Buffalo CSE Department Lines: 28 Distribution: sunyab Message-ID: <9nntiu$1s7$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:29 sunyab.cse.575:39 ------------------------------------------------------------------------- Subject: OFFICE HOURS ------------------------------------------------------------------------- My office hours for Fall 2001 will be: Mondays 10 am - noon Tuesdays 2 - 3:30 pm or by appointment. ------------------------------------------------------------------------- William J. Rapaport Associate Professor of Computer Science & Adjunct Professor of Philosophy Member, Center for Cognitive Science Associate Director, SNePS Research Group (SNeRG) 201 Bell Hall (office: 214 Bell) | work: 716-645-3180 x 112 Dept. of Computer Science & Engineering | fax: 716-645-3464 University at Buffalo (SUNY) | home: 716-636-8625 Buffalo, NY 14260-2000 | rapaport@cse.buffalo.edu ------------------------------------------------------------------------- CSE: http://www.cse.buffalo.edu/ homepage: /~rapaport/ SNeRG: /sneps/ Buffalo Restaurant Guide: /restaurant.guide/ Center for Cognitive Science: http://wings.buffalo.edu/cogsci/ From - Thu Sep 13 12:49:47 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: NEW C.V.A. PASSAGES Date: 12 Sep 2001 20:40:27 GMT Organization: University at Buffalo CSE Department Lines: 11 Distribution: sunyab Message-ID: <9noh7r$cve$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:30 ------------------------------------------------------------------------- Subject: NEW C.V.A. PASSAGES ------------------------------------------------------------------------- I've added some new passages for the CVA project; these are passages that have an indication of the kind of strategy that might be useful for computing the target word's meaning, and have, I believe, been tested with real people (by the author of the article from which I got them). See: http://www.cse.buffalo.edu/~rapaport/676/F01/cva.passages.txt From - Thu Sep 13 12:49:58 2001 Path: acsu.buffalo.edu!joxer.acsu.buffalo.edu!fdl1 From: Frank D Lehouillier Newsgroups: sunyab.cse.676 Subject: definition of quantifiers Date: Thu, 13 Sep 2001 00:04:17 -0400 Organization: University at Buffalo Lines: 56 Distribution: sunyab Message-ID: NNTP-Posting-Host: joxer.acsu.buffalo.edu Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII NNTP-Posting-User: fdl1 Xref: acsu.buffalo.edu sunyab.cse.676:31 There appears to be a serious mistake in both the syntax and semantics of first order predicate calculus as Martins presents it to us regarding the definition of quantifiers. I am going to use _E for the existential quantifier and _A as the universal quantifier in my examples below. In the definition of _A, number 6 on p. 56 he says that a wff _Ax[alpha] is satisfied by an interpretation iff for every individual constant [or individual ground term?] when you substitute it in for x you get a sentence that is satisfied by the interpretation. The problem is that individual constant is not defined. It probably means one of two things either it means the zero place function letters or it means the individuals in the universe of discourse. Both of these are problematic for the definition as it stands. The zero-place function letters position is untenable because there is not a finite set of zero-place functions. Any number of names for individuals might exist. Second it doesn't take into account individuals who aren't picked out by the interpretation of a zero place function letter. This would lead to the syntax itself being incomplete in the sense that it couldn't refer to states of affairs that actually hold in a model. The sentence _Ax[P(x)] would simply mean that the predicate P applies to everything that has a name. There would be no way to consider things that don't have a name in an interpretation. The other position that the term individual constant refers to the universe of discourse suffers from the fact that it would require substituting in elements of the semantics of the language in for part of the syntactic expression. So in interpreting _Ax[_Ey[P(x,y)] as occurs in Problem 10 of the homework, in order to follow definition 6 on p 56 we would have to check whether both _Ey[P(1,y)] and _Ey[P(2,y)] but these sentences are not interpretable because they are not either FoL or the metalanguage. Martins does acknowledge that there is a problem in how he has things written down by the question mark with the alternate terminology in brackets that is in definition 6. My impression about the way this is usually handled is that there is an interpretation of the variables themselves in any given model so that an expression like P(x,y) has an interpretation in the metalanguage something like (following definition 1 on p 53): P(x,y) is true iff (3,4) is in the relation {(1,2), (1,1)}. Definitions 6 would then be revised to something like the following: _Ax[alpha] is true iff you change the interpretation of x to being equal to another member of the universe of discourse then alpha is still true for every member of the universe of discourse that you might change it to. Definition 7 would likewise be changed so that if there were any interpretation you could change the interpretation of x to in order to make alpha true then _Ex[alpha] would be true. This seems to get around the problem although admittedly it feels like cheating because there is something like syntactic manipulation of the interpreting function which is supposed to be in the metalanguage. Please tell me if I've made a mistake somewhere. Frank LeHouillier From - Thu Sep 13 12:50:06 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.575,sunyab.cse.676 Subject: DIGITAL SUMMIT Date: 13 Sep 2001 16:38:43 GMT Organization: University at Buffalo CSE Department Lines: 40 Distribution: sunyab Message-ID: <9nqnej$nt8$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.575:43 sunyab.cse.676:32 ------------------------------------------------------------------------- Subject: DIGITAL SUMMIT ------------------------------------------------------------------------- Date: Thu, 13 Sep 2001 11:49:51 -0400 From: Kristin Herman Subject: Digital Summit Nov 2 and 3 - further information To: PROVOST-FACULTY-LIST@LISTSERV.ACSU.BUFFALO.EDU *******************HOLD THE DATE NOV 2 AND 3******************** Please disseminate to relevant lists outside of UB. A registration flyer is being printed and will be mailed shortly. The University at Buffalo will host Digital Frontier: The Buffalo Summit 2001, from November 2-3, 2001. This summit will examine the far-reaching impacts of digital technology as manifested in the arts and sciences, medicine, education, and various aspects of everyday life. Plenary speakers include Michael Paige, Director of Xerox PARC, Brenda Laurel, pioneer developer of interactive fiction, Jaron Lanier, who coined the term "virtual reality", Steve Mann, inventor of the wearable computer, and Clifford Stoll, author of Silicon Snake Oil. Symposia will explore privacy, virtual reality, telemedicine and distance learning, information overload and retrieval, the gender divide in tech careers, and person:machine interfaces. These sessions will also include demonstrations of new media, digital arts, and applications of advanced computing, as well as extrapolations about likely technological developments of the future. Sponsors include CISCO Systems, Verizon, Niagara Mohawk, Silicon Graphics, Inc., American Psychological Association and others. The website at http://digitalsummit.buffalo.edu provides further information about the plenary speakers, and also online registration. The summit will take place at the Center for the Arts on the north campus of UB. Questions can be directed to Dr. Cusker in the Office of the Vice President for Research at 645-3321 or cusker@research.buffalo.edu. Jaylan S. Turkkan, Ph.D. Vice President for Research From - Fri Sep 14 12:28:47 2001 Path: acsu.buffalo.edu!not-for-mail From: William J Rapaport Newsgroups: sunyab.cse.676 Subject: Re: definition of quantifiers Date: Fri, 14 Sep 2001 09:26:20 -0400 Organization: SUNY Buffalo Computer Science & Engineering Lines: 76 Distribution: sunyab Message-ID: <3BA2057C.7B176BAE@cse.buffalo.edu> References: NNTP-Posting-Host: adara.cse.buffalo.edu Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit NNTP-Posting-User: rapaport X-Mailer: Mozilla 4.78 [en] (X11; U; SunOS 5.8 sun4u) X-Accept-Language: en Xref: acsu.buffalo.edu sunyab.cse.676:33 Here are my initial replies to Frank's comments: Frank D Lehouillier wrote: > In the definition of _A, number 6 on p. 56 he says that a wff Actually, it's not the *def* of _A, but the *semantics* for _A. > > _Ax[alpha] is satisfied by an interpretation iff for every individual > constant [or individual ground term?] when you substitute it in for x you > get a sentence that is satisfied by the interpretation. The problem is > that individual constant is not defined. But it is; it's defined on p. 37, as functions with 0 arguments. > It probably means one of two > things either it means the zero place function letters Right. > or it means the > individuals in the universe of discourse. No; that would be a confusion of syntax with semantics (as you observe below). > > The zero-place function letters position is untenable because there is not > a finite set of zero-place functions. It doesn't have to be finite. It is, however, countable. > Any number of names for individuals > might exist. Why is that a problem? > Second it doesn't take into account individuals who aren't > picked out by the interpretation of a zero place function letter. This > would lead to the syntax itself being incomplete in the sense that it > couldn't refer to states of affairs that actually hold in a model. This is, in fact, the case for real numbers (i.e., there are reals with no names). So, given a semantic domain containing reals, the language for representing it would, indeed, be incomplete. > The > sentence _Ax[P(x)] would simply mean that the predicate P applies to > everything that has a name. There would be no way to consider things that > don't have a name in an interpretation. That is correct; it is not an error! > > The other position that the term individual constant refers to the > universe of discourse suffers from the fact that it would require > substituting in elements of the semantics of the language in for part of > the syntactic expression. Right; that's the confusion I mentioned above. > Martins does acknowledge that there is a problem in how he has things > written down by the question mark with the alternate terminology in > brackets that is in definition 6. There, the question is whether the semantics should be concerned with complex terms. I don't think that it has to be, since the semantics is recursive. When I get some free time later, I'll post the semantics from a standard text on logic, so that you can compare them. -Bill Rapaport From - Wed Sep 19 15:57:04 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: QUANTIFIER SEMANTICS Date: 19 Sep 2001 15:38:23 GMT Organization: University at Buffalo CSE Department Lines: 43 Distribution: sunyab Message-ID: <9oae5f$r3h$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:34 In reply to Frank's queries about the semantics of FOL, here is a slightly different presentation, excerpted from: Mendelson, Elliott (1979), Introduction to Mathematical Logic, Second Edition (New York: D. Van Nostrand), pp. 50ff. ========================================================================= An *interpretation* M consists of a non-empty set D, called the *domain*..., and an assignment to each predicate letter A^n_J [i.e., A, superscript "n", subscript "j"] of an n-place relation ... in D, to each function letter f^n_j of an n-place operation ... in D (i.e., a function from D^n into D), and to each individual constant a_i of some fixed element ... of D. ... [V]ariables are thought of as ranging over the set D. ... Let there be given an interpretation M with domain D. Let S be the set of denumerable sequences of elements of D. We shall define what it means for a sequence s = (b1, b2, ...) in S to satisfy a wff A in M. [First] define a function s* of one argument, with terms as arguments and values in D: 1. If t is x_i, let s*(t) be b_i 2. If t is an indiv. const., then s*(t) is the interpretation of D of this const. 3. If f^n_i is a function letter and f^M is the corresponding operation in D, and t1..tn are terms, then s*(f^n_i (t1..tn)) = f^M (s*(t1)..s*(tn)). ... Now we proceed to the definition of satisfaction: 1. If A is an atomic wff A^n_m (t1..tn) and A^M is the corresponding relation of the interpretation, then the sequence s satisfies A iff A^M (s*(t1)..s*(tn)), i.e., if the n-tuple (s*(t1)..s*(tn)) is in the relation A^M. 2. s satisfies ~A iff s does not satisfy A 3. s satisfies (A -> B) iff either s does not satisfy A or s satisfies B. 4. s satisfies Ax[A] iff every sequence of S that differs from s in at most the ith component satisfies A. From - Thu Sep 20 13:49:46 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: McCarthy, Advice Taker -- ONLINE! Date: 20 Sep 2001 13:19:09 GMT Organization: University at Buffalo CSE Department Lines: 9 Distribution: sunyab Message-ID: <9ocqcd$rcg$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:35 ------------------------------------------------------------------------- Subject: McCarthy, Advice Taker -- ONLINE! ------------------------------------------------------------------------- An early version of McCarthy's Advice Taker paper is online, along with most of his other writings. Take a look at: http://www-formal.stanford.edu/jmc/mcc59.html http://www-formal.stanford.edu/jmc/ From - Thu Sep 20 13:49:56 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: McCARTHY: GETTING TO THE AIRPORT Date: 20 Sep 2001 13:26:07 GMT Organization: University at Buffalo CSE Department Lines: 17 Distribution: sunyab Message-ID: <9ocqpf$rlf$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:36 ------------------------------------------------------------------------- Subject: McCARTHY: GETTING TO THE AIRPORT ------------------------------------------------------------------------- An interesting follow-up article to McCarthy's "Advice Taker" is: Lifschitz, Vladimir; McCain, Norman; Remolina, Emilio; & Tacchella, Armando (2000), "Getting to the Airport: The Oldest Planning Problem in AI", in Jack Minker (ed.), _Logic-Based Artificial Intelligence_ (Boston: Kluwer Academic Publishers), Ch. 7 (pp. 148-165). From - Thu Sep 20 13:50:03 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: CORRECTED ANSWERS TO HW 1 Date: 20 Sep 2001 13:37:21 GMT Organization: University at Buffalo CSE Department Lines: 14 Distribution: sunyab Message-ID: <9ocreh$s0k$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:37 ------------------------------------------------------------------------- Subject: CORRECTED ANSWERS TO HW 1 ------------------------------------------------------------------------- I have corrected the answers to HW 1. See: http://www.cse.buffalo.edu/~rapaport/676/F01/directory.html or http://www.cse.buffalo.edu/~rapaport/676/F01/hw1.answers.html From - Thu Sep 20 13:50:12 2001 Path: acsu.buffalo.edu!pollux.cse.Buffalo.EDU!pvg From: Paul V Gestwicki Newsgroups: sunyab.cse.676 Subject: Equivalency of constants in First Order Logic Date: Thu, 20 Sep 2001 09:56:34 -0400 Organization: University at Buffalo Lines: 48 Distribution: sunyab Message-ID: NNTP-Posting-Host: pollux.cse.buffalo.edu Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII NNTP-Posting-User: pvg Xref: acsu.buffalo.edu sunyab.cse.676:38 Greetings to all Consider a passage like this: "A man jumped, then he sang." Our human minds tell us that 'he' is the man from the first part of the sentence. However, when doing a mechanical transformation, can we assume that they're the same person? e.g.: Jump(m). Sing(m). [I'm using a prologesque approach to FOL syntax.] or would we have to initially use separate symbols and then assert later that they represent the same semantic object? I don't recall seeing a way of asserting equivalence in FOL, which could lead to a problem if our semantic mapping points to the same entity, but the syntactic structure is stuck with separate symbols. The case above is simplified by the existence of a pronoun which generally must refer to something else. What about one like this: "A handsome man jumped. The dapper gent sang." Now can we still assert immediately that they're the same person? My instincts say that we can't, because we would first need to process some background knowledge that a handsome man could also be dapper, and that "gent" often refers to someone who is a handsome man. So a first attempt at translation into FOL could be something like this: Jump(m). Handsome(m). Sing(g). Dapper(g). or Jump(m). Prop(m,handsome). Sing(g). Prop(g,dapper). or whatever representation you prefer. In this interpretation of the sentence, where we don't assume from the grammar that g amd m are the same, then how can we assert their equivalence? This can be solved by using existentailly quantified variables rather than constants to represent the characters, but this doesn't seem to capture the essence of the original sentence (donkeys notwithstanding). Does anyone have any other ideas? My CVA passage has a similar structure to it, and right now I am simply assuming that the natural-language parser can determine that the subject of the sentence has not changed (introducing the word "then" would certainly facilitate this). I realize there's a number of issues involved in this posting, but I've always enjoyed prompting newsgroup discussion. :) Cheers, Paul From - Fri Sep 21 09:08:19 2001 Path: acsu.buffalo.edu!not-for-mail From: William J Rapaport Newsgroups: sunyab.cse.676 Subject: Re: Equivalency of constants in First Order Logic Date: Thu, 20 Sep 2001 13:57:58 -0400 Organization: SUNY Buffalo Computer Science & Engineering Lines: 55 Distribution: sunyab Message-ID: <3BAA2E26.76B865CC@cse.buffalo.edu> References: NNTP-Posting-Host: adara.cse.buffalo.edu Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit NNTP-Posting-User: rapaport X-Mailer: Mozilla 4.78 [en] (X11; U; SunOS 5.8 sun4u) X-Accept-Language: en Xref: acsu.buffalo.edu sunyab.cse.676:39 Paul V Gestwicki wrote: > > Consider a passage like this: > "A man jumped, then he sang." > > Our human minds tell us that 'he' is the man from the first part of the > sentence. However, when doing a mechanical transformation, can we assume > that they're the same person?...or would we have to initially use separate > symbols > and then assert later > that they represent the same semantic object? I don't recall seeing a way > of asserting equivalence in FOL, Well, as you now know, we can easily add "=" to FOL. So, you'd have 2 options for translating the above 2 sentences: Jm ^ Sm or: Jm ^ Sn ^ m=n Most people, using ordinary language, would do the former. If there were several intervening sentences between these 2, so that you weren't entirely sure whether "he" who sang was "the man" who jumped, you might do the latter. > The case above is simplified by the existence of a pronoun which > generally must refer to something else. What about one like this: > "A handsome man jumped. The dapper gent sang." You could do the same, more or less: either: Hm ^ Mm ^ Jm ^ Dm ^ Gm ^ Sm or: Hm ^ Mm ^ Jm ^ Dn ^ Gn ^ Sn ^ m=n You'd need a theory of how to decide when and whether the dapper gent (n) was the handsome man (m). This is the subject of discourse theory withing linguistics (computational or otherwise). > My CVA passage has a similar > structure to it, and right now I am simply assuming that the > natural-language parser can determine that the subject of the sentence has > not changed (introducing the word "then" would certainly facilitate this). You should do what seems "reasonable" in giving your translation, leaving the task of deciding how and when to identify individuals to the parsing program. -Bill From - Fri Sep 21 09:08:28 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: QUERY ON TRANSLATING INTO F.O.L. Date: 20 Sep 2001 19:00:47 GMT Organization: University at Buffalo CSE Department Lines: 64 Distribution: sunyab Message-ID: <9odecv$9fd$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:40 A student writes: > I'm working on translating the medical document to first-order logic. I > have a question about the translation--it would seem logical that for > each > sentence, I could create a new, equivalent first-order logic statement. > But, to encompass the entire document, wouldn't it have to be contained > in > one (rather large) statement? I ask because otherwise, I would have to > (for example) represent Fred in each sentence by first saying > > Ex[Fred(x)...] > > Whereas it seems more true to the original English if I only have to > state > that once, at the beginning of one long statement. > > Hopefully that made sense--any suggestions? If you were going to translate into SNePS, it would be easy to resolve this problem, since the translation of each sentence would be a network that would automatically be linked to the network for all preceding (and, eventually, succeeding) sentences, via sharing of common nodes. Using FOL, it's not quite as obvious how to handle this, since FOL isn't strictly speaking a KR language. But the solution, I think, is related to my comments on Paul's question. Basically, you're dealing with discourse, not isolated sentences. Hence, the representation of any given sentence must be considered in the light of your representations of the previous sentences. So, if you have a text like this: Fred is tall. He is handsome. you could either represent it as 2 separate sentences: Tf. Hf. or as a conjunction: Tf ^ Hf. However, we can assume, for present purposes, that these are equivalent, i.e., that separate sentences are implicitly conjoined. Note, by the way, that you don't have to do the following: Ex[Fred(x) ^ Tx ^ Hx] and then somehow have to conjoin in the rest of the sentences to ensure that they all talk about Fred. Of course, you can't do this either: ExFred(x). Tx. Hx. since the "x"s in the last 2 sentences are unbound (recall our discussion of the donkey sentence). Instead, you can use Skolem constants (which, by the way, is essentially how it's done in SNePS): Fred(f). Tf. Hf. Let me know if you have further questions. From - Fri Sep 21 09:08:52 2001 Path: acsu.buffalo.edu!mes6 From: Michael E Schiller Newsgroups: sunyab.cse.676 Subject: FOL Numeric Values? Date: Fri, 21 Sep 2001 00:47:26 -0400 Organization: University at Buffalo Lines: 22 Distribution: sunyab Message-ID: NNTP-Posting-Host: pollux.cse.buffalo.edu Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII NNTP-Posting-User: mes6 Originator: mes6@pollux.cse.Buffalo.EDU Xref: acsu.buffalo.edu sunyab.cse.676:41 Here's a question: How can one deal with numeric values in first order logic? I see SNePS seems to deal with it using "properties," where you could say something along the lines of has-property(Bob,height,6) ...where height is being measured in feet. Can we do something similar in first order logic? For example, would something like Height(b,6) (where "b" is representing Bob) be valid? Or perhaps Has(b,h,6) (where "b" is Bob and "h" is height) ? Thanks! Mike S From - Fri Sep 21 12:36:52 2001 Path: acsu.buffalo.edu!not-for-mail From: William J Rapaport Newsgroups: sunyab.cse.676 Subject: Re: FOL Numeric Values? Date: Fri, 21 Sep 2001 09:11:01 -0400 Organization: SUNY Buffalo Computer Science & Engineering Lines: 23 Distribution: sunyab Message-ID: <3BAB3C65.86787BB6@cse.buffalo.edu> References: NNTP-Posting-Host: adara.cse.buffalo.edu Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit NNTP-Posting-User: rapaport X-Mailer: Mozilla 4.78 [en] (X11; U; SunOS 5.8 sun4u) X-Accept-Language: en Xref: acsu.buffalo.edu sunyab.cse.676:42 Michael E Schiller wrote: > Here's a question: > How can one deal with numeric values in first order logic? The simplest way is probably to treat them as individual constants. > I see SNePS > seems to deal with it using "properties," Not quite, but we'll discuss how SNePS handles them (poorly!) when we discuss SNePS in lecture. > would something like > > Height(b,6) (where "b" is representing Bob) > > be valid? Or perhaps > > Has(b,h,6) (where "b" is Bob and "h" is height) Or use function symbols: height(bob) = 6 From - Wed Sep 26 09:19:29 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: TERM PROJECT Date: 25 Sep 2001 13:40:05 GMT Organization: University at Buffalo CSE Department Lines: 29 Distribution: sunyab Message-ID: <9oq1fl$fab$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:43 ------------------------------------------------------------------------- Subject: TERM PROJECT ------------------------------------------------------------------------- Here's an interesting question: What should you do if you believe that you will have completed the representation of your passage very early in the semester? Does that mean that you are finished with your term project? No! After you have completely represented (to your satisfaction) your passage *plus* all relevant background knowledge, in your favorite KR language, you should do one or more of the following: 1. If your favorite KR language is not SNePS, then try representing your passage and background knowledge in SNePS. 2. If you have represented your passage plus background knowledge in SNePS by hand, then implement it in SNePS. 3. Independently of 1 and 2 above, if you are representing a CVA passage, then represent *more* passages with the same unknown word (together with *more* background knowledge), or, if you are representing one of the hospital discharge summaries, then you could either represent another one (I will post a list of who is doing what) or try to represent further background knowledge that might be useful for the "translation into plain English" task. From - Wed Sep 26 09:19:51 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: TERM PROJECT TOPICS Date: 25 Sep 2001 13:51:30 GMT Organization: University at Buffalo CSE Department Lines: 37 Distribution: sunyab Message-ID: <9oq252$fkl$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:44 ------------------------------------------------------------------------- Subject: TERM PROJECT TOPICS ------------------------------------------------------------------------- Here is a list of current choices for your term projects: ------------------------------------------------------------------------- CVA passages: ------------------------------------------------------------------------- Marc K Broklawski "proliferate" Chienchih Chi "sedative" Paul V Gestwicki "oam" Valerie Yakich "proximity" Here's an interesting passage that the 4 of you might collaborate on: "The oam proliferated in the proximity of the fire, acting as a sedative." :-) ------------------------------------------------------------------------- HDS passages: ------------------------------------------------------------------------- Frank LeHouillier Shamel Robert J Milewski Teron Michael Schiller Pablo Rajeev M Sood Nicholas ------------------------------------------------------------------------- Passage not yet chosen (please do so NOW!) ------------------------------------------------------------------------- Kazuhiro Kawachi From - Wed Sep 26 09:19:56 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: UPDATED PROJECT LIST Date: 26 Sep 2001 13:16:20 GMT Organization: University at Buffalo CSE Department Lines: 23 Distribution: sunyab Message-ID: <9oskf4$ss1$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:45 ------------------------------------------------------------------------- Subject: UPDATED PROJECT LIST ------------------------------------------------------------------------- CVA passages: ------------------------------------------------------------------------- Marc K Broklawski "proliferate" Chienchih Chi "sedative" Paul V Gestwicki "oam" Kazuhiro Kawachi "taciturn" Valerie Yakich "proximity" ========================================================================= HDS passages: ------------------------------------------------------------------------- Frank LeHouillier Shamel Robert J Milewski Teron Michael Schiller Pablo Rajeev M Sood Nicholas From - Tue Oct 2 08:46:32 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: SCHEDULE OF FRIDAY "MASTER CLASSES" Date: 1 Oct 2001 14:37:31 GMT Organization: University at Buffalo CSE Department Lines: 24 Distribution: sunyab Message-ID: <9p9v3b$ije$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:46 ------------------------------------------------------------------------- Subject: SCHEDULE OF FRIDAY "MASTER CLASSES" ------------------------------------------------------------------------- Here is a schedule of our Friday "master classes", so that you will have a chance to prepare your presentation. Date Person Topic Status ------------------------------------------------------------------------- Sep 21 Gestwicki CVA: "oam" done Sep 28 LeHouillier HDS: Shamel done Oct 5 Broklawski CVA: "proliferate" Oct 12 Sood HDS: Nicholas ------------------------------------------------------------------------- *** NOTE: TERM PROJECT PROGRESS REPORTS DUE OCT 12 *** ------------------------------------------------------------------------- Oct 19 Yakich CVA: "proximity" Oct 26 Milewski HDS: Teron Nov 2 Chi CVA: "sedative" Nov 9 Schiller HDS: Pablo Nov 16 Kawachi CVA: "taciturn" Nov 30 open From - Tue Oct 2 08:46:49 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: HOSPITAL DISCHARGE SUMMARIES Date: 1 Oct 2001 15:09:21 GMT Organization: University at Buffalo CSE Department Lines: 13 Distribution: sunyab Message-ID: <9pa0v1$jf9$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:47 ------------------------------------------------------------------------- Subject: HOSPITAL DISCHARGE SUMMARIES ------------------------------------------------------------------------- I have reorganized the PDF files in Kevin Stevens's project report and list of slides, in PDF format at: http://www.cse.buffalo.edu/~rapaport/676/F01/MEDNLP to make them easier to use. Those of you working on this project should try to use/adapt/modify Kevin's case frames wherever possible. From - Tue Oct 2 08:47:11 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: KNOWLEDGE REPRESENTATION AND SCIENCE FICTION Date: 1 Oct 2001 15:12:39 GMT Organization: University at Buffalo CSE Department Lines: 11 Distribution: sunyab Message-ID: <9pa157$jic$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:48 ------------------------------------------------------------------------- Subject: KNOWLEDGE REPRESENTATION AND SCIENCE FICTION ------------------------------------------------------------------------- You might be interested in the following science-fiction story by none other than John McCarthy. Full appreciation of it depends heavily on some knowledge of knowledge representation techniques. McCarthy, John (2001), "The Robot and the Baby" http://www-formal.stanford.edu/jmc/robotandbaby.html From - Tue Oct 2 08:47:25 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: CSE Colloquium: Alberto Mendelzon, Data (and Links) on the Web Date: 1 Oct 2001 16:37:22 GMT Organization: University at Buffalo CSE Department Lines: 26 Distribution: sunyab Message-ID: <9pa642$lo8$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:49 ------------------------------------------------------------------------- A colloquium of possible relevance to knowledge representation: ------------------------------------------------------------------------- TITLE: Data (and Links) on the Web SPEAKER: Alberto Mendelzon, University of Toronto TIME: Friday, October 5, 3:00pm LOCATION: Bell 224 ABSTRACT: In the excitement of extending database technology such as semistructured models and languages to the Web, the importance of links between documents has somehow been left behind. We will present two projects that keep links front and center: the WebSQL/WebOQL query languages and the TOPIC system for computing page reputations. WebSQL and WebOQL are query languages intended to serve as high-level tools for building Web-based information systems in the same way that SQL is used for building traditional information systems. TOPIC is a method for analyzing incoming links to a page in order to determine what are the topics that this page is best known for on the Web. CSE colloquium schedule: http://www.cse.buffalo.edu/colloq/Fall2001-Colloquium.html From - Tue Oct 2 08:47:33 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: HOW TO LOG INTO CSE MACHINES TO RUN SNePS Date: 1 Oct 2001 20:10:47 GMT Organization: University at Buffalo CSE Department Lines: 19 Distribution: sunyab Message-ID: <9paik7$qv4$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:50 ------------------------------------------------------------------------- Subject: HOW TO LOG INTO CSE MACHINES FROM UBUnix TO RUN SNePS ------------------------------------------------------------------------- 1. Log onto UBUnix (either on campus or from home) 2. At the UBUnix prompt, type: rlogin pollux.cse.buffalo.edu 3. Log onto pollux Your user name and password should be the same as for UBUnix 4. Run SNePS [as described in the tutorial] 5. Log off pollux 6. Log off UBUnix From - Wed Oct 3 09:34:09 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: AI Meets Natural Stupidity Date: 2 Oct 2001 19:47:40 GMT Organization: University at Buffalo CSE Department Lines: 7 Distribution: sunyab Message-ID: <9pd5ks$3ps$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:51 ------------------------------------------------------------------------- Subject: AI Meets Natural Stupidity ------------------------------------------------------------------------- This paper is now on reserve; go to Bison -> Course Reserve, enter "CSE676" (no spaces) for Course No., and start the search. It's the first (and/or maybe second) item on the list. From - Tue Oct 9 09:23:40 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: AI Meets Natural Stupidity Date: 8 Oct 2001 13:53:22 GMT Organization: University at Buffalo CSE Department Lines: 9 Distribution: sunyab Message-ID: <9psb4i$g16$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:54 ------------------------------------------------------------------------- Subject: AI Meets Natural Stupidity ------------------------------------------------------------------------- This paper is now on reserve; go to Bison -> Course Reserve, enter "CSE676" (no spaces) for Course No., and start the search. It's the first (and/or maybe second) item on the list. From - Tue Oct 9 09:23:51 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: 3 PAPERS ON INTENSIONALITY IN SNePS Date: 8 Oct 2001 14:17:56 GMT Organization: University at Buffalo CSE Department Lines: 8 Distribution: sunyab Message-ID: <9pscik$gk8$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:55 ------------------------------------------------------------------------- Subject: 3 PAPERS ON INTENSIONALITY IN SNePS ------------------------------------------------------------------------- I've posted the references to the 3 papers on intensionality in SNePS that I mentioned in lecture to the directory of documents. From - Wed Oct 10 09:04:56 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: BARCAN FORMULA Date: 9 Oct 2001 13:29:37 GMT Organization: University at Buffalo CSE Department Lines: 9 Distribution: sunyab Message-ID: <9puu41$pch$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:56 ------------------------------------------------------------------------- Subject: BARCAN FORMULA ------------------------------------------------------------------------- I have put a link to a webpage on modal logic generally and the Barcan formula in particular on the Directory of Documents. http://www.cse.buffalo.edu/~rapaport/676/F01/directory.html From - Wed Oct 10 09:05:09 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: CORRESPONDENCE CONTINUUM Date: 9 Oct 2001 14:14:35 GMT Organization: University at Buffalo CSE Department Lines: 14 Distribution: sunyab Message-ID: <9pv0ob$qc7$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:57 ------------------------------------------------------------------------- Subject: CORRESPONDENCE CONTINUUM ------------------------------------------------------------------------- The following paper: Smith, Brian Cantwell (1987), "The Correspondence Continuum", Report CSLI-87-71 (Stanford, CA: Center for the Study of Language and Information). is now online. Go to Bison -> Course Reserve -> CSE676 or our Directory of Documents. From - Wed Oct 10 09:05:21 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: TERM PROJECT PROGRESS REPORT Date: 10 Oct 2001 12:54:28 GMT Organization: University at Buffalo CSE Department Lines: 37 Distribution: sunyab Message-ID: <9q1ge4$5lg$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:58 ------------------------------------------------------------------------- Subject: TERM PROJECT PROGRESS REPORT ------------------------------------------------------------------------- As you know, the term-project progress report is due on Friday. A student writes: > Do you want the progress report as a formal report? I remember > you saying that writing syntax and semantics were sufficient. Is there > anything else you would like included? You should provide at least the following: 1. A copy of your passage. 2. The name of the knowledge representation language you are using. 3. The syntax and semantics of the representations you have developed so far 4. The representations of the passage (and any background knowledge, if you've gotten that far; see below) that you have so far. I strongly suggest, so that you avoid a depth-first traversal (see below), that you completely represent the passage first, and *then* turn your attention to representing background knowledge. You probably should, however, make some notes about what background knowledge you will eventually need. (For those of you not familiar with depth-first traversal, the point I was trying to make above is that if you have a sequence of tasks to perform, each of which has a sequence of subtasks (and so on, recursively), you are usually better off performing the tasks in sequence (i.e., breadth-first), then performing the subtasks in sequence, etc. Otherwise, if you first perform task 1, then perform the first of its subtasks, *then* perform the first of that subtask's subtasks, etc. (i.e., depth-first), you run the rist of never getting back to doing task 2.) From - Thu Oct 11 08:55:16 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: SYLLABUS UPDATED, 10-10-01 Date: 10 Oct 2001 13:19:30 GMT Organization: University at Buffalo CSE Department Lines: 11 Distribution: sunyab Message-ID: <9q1ht2$6bp$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:59 ------------------------------------------------------------------------- Subject: SYLLABUS UPDATED, 10-10-01 ------------------------------------------------------------------------- I have updated the syllabus to: 1. fine-tune the "tentative" schedule 2. fine-tune the readings for the unit on non-monotonic logics See: http://www.cse.buffalo.edu/~rapaport/676/F01/syl.html#dates From - Thu Oct 11 08:55:22 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: EPISTEMIC LOGIC Date: 10 Oct 2001 13:59:50 GMT Organization: University at Buffalo CSE Department Lines: 6 Distribution: sunyab Message-ID: <9q1k8m$79s$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:60 ------------------------------------------------------------------------- Subject: EPISTEMIC LOGIC ------------------------------------------------------------------------- I've updated the directory of documents to conveniently list in one place various references on epistemic logic. From - Mon Oct 15 12:22:23 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: WRITING GUIDELINES Date: 12 Oct 2001 18:06:42 GMT Organization: University at Buffalo CSE Department Lines: 12 Distribution: sunyab Message-ID: <9q7bfi$fsl$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:61 ------------------------------------------------------------------------- Subject: WRITING GUIDELINES ------------------------------------------------------------------------- A reminder for those of you writing term projects: All programming and research reports must follow my on-line writing guidelines, at: http://www.cse.buffalo.edu/~rapaport/typing.info.pdf From - Mon Oct 15 12:22:33 2001 Path: acsu.buffalo.edu!pollux.cse.Buffalo.EDU!pvg From: Paul V Gestwicki Newsgroups: sunyab.cse.676 Subject: Mars Mission: We Don't Trust AI Date: Fri, 12 Oct 2001 16:18:36 -0400 Organization: University at Buffalo Lines: 27 Distribution: sunyab Message-ID: NNTP-Posting-Host: pollux.cse.buffalo.edu Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII NNTP-Posting-User: pvg Xref: acsu.buffalo.edu sunyab.cse.676:62 Here's an amusing story that I was reminded of when we discussed expert systems in today's class. About two years ago, some Comp Sci friends and I went to the Buffalo Science Center to attend a presentation given by one of the engineers of the Mars Pathfinder mission. I'm afraid I don't remember his name or capacity offhand, but he was associated with some university in New York. He spoke a bit about the problems of remote controlling a rover, especially the problem of there being only a limited number of hours each day in which communication can take place. At the reception afterwards, my friends and I approached him and asked whether AI was to be used to make the rover(s) autonomous. (The friends I was with are programmers by trade and robotics hobbyists.) His response was that the engineers and scientists who created the rover and the mission would never trust their project to the mysterious ways of computer scientists. Now here is a socialogical hurdle we must cross as well as the scientific ones! For those of you who were involved in mine and Rob's conversation earlier this semester, I use this example as proof that we are higher up in the Hierarchy of Geekdom than they are. :) Enjoy the weekend, Paul From - Mon Oct 15 16:25:59 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: MCDERMOTT'S FALLACY Date: 15 Oct 2001 19:55:14 GMT Organization: University at Buffalo CSE Department Lines: 18 Distribution: sunyab Message-ID: <9qfev2$baj$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:63 ------------------------------------------------------------------------- Subject: MCDERMOTT'S FALLACY ------------------------------------------------------------------------- A student writes: > I need a little bit of clarification as to what exactly it means to be > falling victim to McDermott's Fallacy. As an example, I called one node > in my SNePS representation "thisvisit". As is, with no explanation to > explain what "thisvisit" is, I can see it as an example of the Fallacy. > My question is, to fix this, would it suffice to just have an explanation > of what "thisvisit" represents? It would suffice if the explanation was expressed in SNePS. So, you'll need rules that explain that "thisvisit" is a visit, that it is the current visit, maybe also rules that explain what a visit is, etc. Whatever you would provide in English, you should represent in SNePS as part of the background knowledge. From - Wed Oct 24 12:42:16 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: BIRDS GOTTA FLY Date: 24 Oct 2001 16:00:19 GMT Organization: University at Buffalo CSE Department Lines: 9 Distribution: sunyab Message-ID: <9r6oij$cba$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:64 ------------------------------------------------------------------------- Subject: BIRDS GOTTA FLY ------------------------------------------------------------------------- Check out the default-reasoning cartoon on the updated Directory of Documents. http://www.cse.buffalo.edu/~rapaport/676/F01/directory.html From - Thu Oct 25 09:04:58 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: SNeBR DEMO Date: 24 Oct 2001 20:20:44 GMT Organization: University at Buffalo CSE Department Lines: 57 Distribution: sunyab Message-ID: <9r77qs$jm4$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:65 ------------------------------------------------------------------------- Subject: SNeBR DEMO ------------------------------------------------------------------------- If you want to do the SNeBR demo on your own, you might find the following instructions from Fran Johnson of use: Date: Tue, 23 Oct 2001 14:51:38 -0400 (EDT) From: Frances L Johnson cc: "William J. Rapaport" Subject: Pegasus demo You have 2 choices regarding the snepslog pegasus demo... 3, actually, if you include NOT doing it :-) 1) You can run the demo interactively: a) run the demo b) at end, *YOU* cut & paste the last query & hit return c) *YOU* interact with the manual SNeBR interface to remove the inconsistency OR... 2) you run the preset demo that I made to get you through the interface, then scroll back up to show what it did. Instructions for both options (1 and 2) are given below. Fran ---------------------------- INSTRUCTIONS FOR 1) load sneps at lisp prompt: (snepslog) at snepslog prompt: demo choose demo #4: 4 ... when demo fully ends: flies(pegasus)? snebr activated! best answers are: r 2 d q n n should result in "FLIES(PEGASUS)" lastly... at snepslog prompt: list-asserted-wffs to return to lisp... at snepslog prompt: lisp ------------------------------ INSTRUCTIONS FOR 2) load sneps at lisp prompt: (snepslog) at snepslog prompt: demo "/projects/flj/pegasus-full.snlog" av when done, scroll up to see the SNeBR interaction From - Thu Oct 25 09:10:25 2001 Path: acsu.buffalo.edu!not-for-mail From: William J Rapaport Newsgroups: sunyab.cse.676 Subject: Re: MIDSEMESTER COURSE EVALUATION Date: Thu, 25 Oct 2001 09:09:21 -0400 Organization: SUNY Buffalo Computer Science & Engineering Lines: 12 Distribution: sunyab Message-ID: <3BD80F01.2F9E471@cse.buffalo.edu> References: <9r792b$k6v$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit NNTP-Posting-User: rapaport X-Mailer: Mozilla 4.78 [en] (X11; U; SunOS 5.8 sun4u) X-Accept-Language: en Xref: acsu.buffalo.edu sunyab.cse.676:67 "William J. Rapaport" wrote: > ... read some > other SNePS papers from the SNeRG bibliography at > http://www.cse.buffalo.edu//sneps/Bibliography/ Although that URL *will* work, there's an extra "/" in it; it really should have been: http://www.cse.buffalo.edu/sneps/Bibliography/ From - Thu Oct 25 12:40:40 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: TERM-PROJECT GRADING CRITERIA Date: 25 Oct 2001 13:27:20 GMT Organization: University at Buffalo CSE Department Lines: 11 Distribution: sunyab Message-ID: <9r93vo$l0l$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:68 ------------------------------------------------------------------------- Subject: TERM-PROJECT GRADING CRITERIA ------------------------------------------------------------------------- A webpage on the criteria I'll be using for grading your term projects is at: http://www.cse.buffalo.edu/~rapaport/676/F01/termprojectgrading.html (and has been listed on the Directory of Documents). From - Thu Oct 25 12:40:57 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: DIRECTORY OF DOCUMENTS RE-ORGANIZED Date: 25 Oct 2001 14:43:14 GMT Organization: University at Buffalo CSE Department Lines: 11 Distribution: sunyab Message-ID: <9r98e2$mvh$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:69 ------------------------------------------------------------------------- Subject: DIRECTORY OF DOCUMENTS RE-ORGANIZED ------------------------------------------------------------------------- In response to popular demand (including my own), I have completely re-organized the Directory of Documents. (Hopefully, all the material that used to be on it is still retrievable from it.) Check it out at: http://www.cse.buffalo.edu/~rapaport/676/F01/directory.html From - Thu Oct 25 14:39:16 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: MIDSEMESTER COURSE EVALUATION Date: 24 Oct 2001 20:41:47 GMT Organization: University at Buffalo CSE Department Lines: 90 Distribution: sunyab Message-ID: <9r792b$k6v$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:66 ------------------------------------------------------------------------- Subject: MIDSEMESTER COURSE EVALUATION ------------------------------------------------------------------------- Thanks for your comments. For the most part, you seem happy with the course (:-) but have a few suggestions for improvement, which I'll comment on below. (Some of your suggestions, as you note, are out of my control, such as the length of the class, the location, etc., so I won't comment on those.) If, at any time, you have other suggestions for improvement, please let me know. ========================================================================= 1. Some of you wanted clarification on how your grade will be determined. According to the syllabus, "Your final course grade will be a weighted average (probably 50-50) of (1) your class attendance, class participation, and homeworks, and (2) your grade on the project." Now, I did think that there might be more homeworks than there have in fact been. (I've been somewhat disappointed in the choice of exercises at the end of the Martins chapters.) In any case, homeworks aren't graded. What this means, essentially, is that if you attend virtually all classes and actively participate (which, so far, is true of all of you), then your "class" grade will be A. So your final grade will be the average of your project grade and A. I will post the criteria I will use to grade your project. ------------------------------------------------------------------------- 2. Some of you would like *more* work: quizzes or written summaries of the readings, for example. I could do that, but quizzes will just limit the amount of time we have to cover material, and my experience with written summaries is that they are either poorly done or perfunctory. I'm not sure anyone gets anything out of them. You are grad students, and this is an advanced grad course. Many of these papers are "classics" and well worth reading. You should feel guilty if you don't read them :-) ------------------------------------------------------------------------- 3. Some of you are unhappy with having to go to the library to access Brachman & Levesque and Ginsberg. Me, too. But I can't put those papers on line; the library will know that they all come from the same book, and they just won't do it. ------------------------------------------------------------------------- 4. Some of you think that the Directory of Documents is cluttered. Yeah; bummer! I find that I can't find things easily on it, either! When I get a chance, I'll try to clean it up a bit. ------------------------------------------------------------------------- 5. Some of you wanted more example implementations of the concepts, or more applications of the different KR languages to problems relevant to the projects. I'll certainly keep this in mind as I present material, but perhaps the best way to do this is on Friday master classes when we are looking at actual representational issues. ------------------------------------------------------------------------- 6. Some of you suggested that I give you more time to get up the nerve to ask questions when I say "any questions?". Funny thing about that: As chair of the CSE Teaching Quality Committee, I often tell my colleagues to do just that (in particular, to count to 10). Maybe I need to count to 15 :-) More seriously, if you do have questions but don't get a chance to ask them in class, post them to the newsgroup or mail them to me. If I feel that they are of general interest, I'll post them anonymously to the newsgroup and reply there. If you prefer that I *not* do that with your question, just let me know. ------------------------------------------------------------------------- 7. Some of you want to know more about SNePS. I'll try, but there's only so much time in the semester. Join our SNePS Research Group (SNeRG) on Fridays, 9:30-noon, Bell 224, or read some other SNePS papers from the SNeRG bibliography at http://www.cse.buffalo.edu//sneps/Bibliography/ ------------------------------------------------------------------------- From - Thu Oct 25 14:39:22 2001 Path: acsu.buffalo.edu!not-for-mail From: William J Rapaport Newsgroups: sunyab.cse.676 Subject: Re: MIDSEMESTER COURSE EVALUATION Date: Thu, 25 Oct 2001 09:09:21 -0400 Organization: SUNY Buffalo Computer Science & Engineering Lines: 12 Distribution: sunyab Message-ID: <3BD80F01.2F9E471@cse.buffalo.edu> References: <9r792b$k6v$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit NNTP-Posting-User: rapaport X-Mailer: Mozilla 4.78 [en] (X11; U; SunOS 5.8 sun4u) X-Accept-Language: en Xref: acsu.buffalo.edu sunyab.cse.676:67 "William J. Rapaport" wrote: > ... read some > other SNePS papers from the SNeRG bibliography at > http://www.cse.buffalo.edu//sneps/Bibliography/ Although that URL *will* work, there's an extra "/" in it; it really should have been: http://www.cse.buffalo.edu/sneps/Bibliography/ From - Fri Oct 26 09:13:39 2001 Path: acsu.buffalo.edu!not-for-mail From: Robert Milewski Newsgroups: sunyab.cse.676 Subject: charting Date: Thu, 25 Oct 2001 23:18:47 -0400 Organization: University at Buffalo Lines: 6 Distribution: sunyab Message-ID: <9rak7n$eke$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: resnet70-154.resnet.buffalo.edu Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.2) Gecko/20010726 Netscape6/6.1 X-Accept-Language: en-us Xref: acsu.buffalo.edu sunyab.cse.676:70 If you are having issues with efficient diagraming... I strongly recommend the program "Dia" - it is included with Gnome (e.g. using Red Hat Linux) - amazing ! Save in a png or ps when done and use gimp to convert it to a jpg. From - Fri Oct 26 11:17:30 2001 Path: acsu.buffalo.edu!not-for-mail From: William J Rapaport Newsgroups: sunyab.cse.676 Subject: Re: charting Date: Fri, 26 Oct 2001 09:15:28 -0400 Organization: SUNY Buffalo Computer Science & Engineering Lines: 19 Distribution: sunyab Message-ID: <3BD961F0.92BD7E5@cse.buffalo.edu> References: <9rak7n$eke$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit NNTP-Posting-User: rapaport X-Mailer: Mozilla 4.78 [en] (X11; U; SunOS 5.8 sun4u) X-Accept-Language: en Xref: acsu.buffalo.edu sunyab.cse.676:71 Robert Milewski wrote: > If you are having issues with efficient diagraming... > I strongly recommend the program "Dia" - it is included > with Gnome (e.g. using Red Hat Linux) - amazing ! > Save in a png or ps when done and use gimp to convert > it to a jpg. If I understand the essence of your message, this may be of potential interest to those of us who draw SNePS nets, but a few items need to be translated for those of us who are not as computer-savvy as you: Gnome png gimp Is this available on the CSE Unix machines? From - Fri Oct 26 14:59:47 2001 Path: acsu.buffalo.edu!not-for-mail From: Robert Milewski Newsgroups: sunyab.cse.676 Subject: Re: charting Date: Fri, 26 Oct 2001 12:12:02 -0400 Organization: University at Buffalo Lines: 142 Distribution: sunyab Message-ID: <3BD98B52.95D3941F@cse.buffalo.edu> References: <9rak7n$eke$1@prometheus.acsu.buffalo.edu> <3BD961F0.92BD7E5@cse.buffalo.edu> NNTP-Posting-Host: trifid.cedar.buffalo.edu Mime-Version: 1.0 Content-Type: multipart/alternative; boundary="------------CBAB5B1D7B4D8FAA39735C56" X-Mailer: Mozilla 4.76 [en] (X11; U; SunOS 5.7 sun4u) X-Accept-Language: en Xref: acsu.buffalo.edu sunyab.cse.676:72 --------------CBAB5B1D7B4D8FAA39735C56 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit www.gnome.org www.gimp.org PNG is an image format that most graphical programs support (e.g. gimp, photoshop, etc...) type 'gimp' in unix or linux - (gimp == photoshop ^ freehand) you can use it to draw or to convert image file types. gnome is the desktop manager provided with the linux operating system (KDE is its competitor). the window manager controls all of the windows properties. the desktop manager controls all of your icons, drag n drop, file management, menu system,and nowadays, a host of utilities in many categories. picture the "start" bar, desktop and windows explorer in windows - well the equivalent in linux to that concept is gnome. - except, it will be a picture of a foot instead of the word "start". is dia provided for unix ? well typically, the source code for *nix applications can be compiled for most other *nix programs. as for binaries - i dunno. but i do know red hat linux, solaris, mandrake linux, debian linux - typically are the top binaries to download - and there are usually very minor issues with compiling source (e.g. dL a new library or something) www.gnome.org might have more information for a solaris version - or a link to the dia website (whatever that may be). to all who don't use linux - i strongly recommend it - it is a very nice bridge between windows and unix - without losing the technical superiority of *nix and maintaining, to a good degree, the user friendliness. William J Rapaport wrote: > Robert Milewski wrote: > > > If you are having issues with efficient diagraming... > > I strongly recommend the program "Dia" - it is included > > with Gnome (e.g. using Red Hat Linux) - amazing ! > > Save in a png or ps when done and use gimp to convert > > it to a jpg. > > If I understand the essence of your message, this may be > of potential interest to those of us who draw SNePS nets, > but a few items need to be translated for those of us who > are not as computer-savvy as you: > > Gnome > png > gimp > > Is this available on the CSE Unix machines? -- Robert Milewski (716) 645-6164 x507 Research Scientist Center of Excellence for Document Analysis and Recognition State University of New York at Buffalo --------------CBAB5B1D7B4D8FAA39735C56 Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit www.gnome.org
www.gimp.org

PNG is an image format that most graphical programs support
(e.g. gimp, photoshop, etc...)

type 'gimp' in unix or linux - (gimp == photoshop ^ freehand)
you can use it to draw or to convert image file types.

gnome is the desktop manager provided with the linux operating system (KDE is its competitor).
the window manager controls all of the windows properties.
the desktop manager controls all of your icons, drag n drop, file management, menu system,and nowadays, a host of utilities in many categories.

picture the "start" bar, desktop and windows explorer in windows - well the equivalent in linux to that concept is gnome. - except, it will be a picture of a foot instead of the word "start".

is dia provided for unix ?
well typically, the source code for *nix applications can be compiled for most other
*nix programs. as for binaries - i dunno. but i do know red hat linux, solaris, mandrake
linux, debian linux - typically are the top binaries to download - and there are usually
very minor issues with compiling source (e.g. dL a new library or something)
www.gnome.org might have more information for a solaris version - or a link to
the dia website (whatever that may be).

to all who don't use linux - i strongly recommend it - it is a very nice bridge between
windows and unix - without losing the technical superiority of *nix and maintaining,
to a good degree, the user friendliness.
 
 

William J Rapaport wrote:

Robert Milewski wrote:

> If you are having issues with efficient diagraming...
> I strongly recommend the program "Dia" - it is included
> with Gnome (e.g. using Red Hat Linux) - amazing !
> Save in a png or ps when done and use gimp to convert
> it to a jpg.

If I understand the essence of your message, this may be
of potential interest to those of us who draw SNePS nets,
but a few items need to be translated for those of us who
are not as computer-savvy as you:

Gnome
png
gimp

Is this available on the CSE Unix machines?

-- 



Robert Milewski  (716) 645-6164 x507
Research Scientist
Center of Excellence for Document Analysis and Recognition
State University of New York at Buffalo
  --------------CBAB5B1D7B4D8FAA39735C56-- From - Tue Oct 30 09:58:20 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: HW #3 Date: 30 Oct 2001 14:42:08 GMT Organization: University at Buffalo CSE Department Lines: 40 Distribution: sunyab Message-ID: <9rme80$51e$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:73 ------------------------------------------------------------------------- Subject: HW #3 ------------------------------------------------------------------------- A student writes: > I just wanted to be sure that I am correct in understanding homework #3. > The problem as stated is to "represent the statements." A simple > representation of statements is straightforward, but some of the > statements are loaded with extra meaning. > > Here's the passage, for reference: > > Anna knows that John bought a dog. The dog is a German Shepherd. > John paid 300 dollars for the dog. Anna also knows that Stu sold his > dog. Anna believes that John bought Stu's dog. > > The last statement, for example, has some hidden meanings. We can say > that "John bought Stu's dog", but in this context what is conveyed (in my > reading of the passage) is actually the proposition "the dog that John > bought is Stu's dog"; there is a connection between this proposition and > the earlier one. This would introduce some EQUIV propositiongs which are > not explicitly in the passage. I don't know if that's going beyond the > scope of the problem, trying to patch together the loose or implicit > references. I think the simplest solution is to use "equiv" arcs, but there are many other ways it could be done. It's your choice, since what you're actually representing is your understanding of the passage. > Unless I hear otherwise, I will proceed with a "simple" translation of > the statements and include in my paper a discussion of the deeper issues. That's always the best thing to do! > Also, the reorganized directory o' documents is much easier to navigate > -- thanks! You're welcome! From - Thu Nov 1 12:51:57 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: FRAME PROBLEM Date: 1 Nov 2001 14:28:51 GMT Organization: University at Buffalo CSE Department Lines: 40 Distribution: sunyab Message-ID: <9rrm73$ac3$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:74 ------------------------------------------------------------------------- Subject: FRAME PROBLEM ------------------------------------------------------------------------- First, I've fixed the "bug" we noticed in my frame-problem handout; there was, indeed, a redundant "result". The corrected version is at: http://www.cse.buffalo.edu/~rapaport/676/F01/frameprob.html Second, a student writes: > This might seem like a basic question, but I was a little bit confused in > class today. In our discussions of the Frame Problem, we went through > some examples of it. What I think I may have missed is exactly what the > "Problem" is--perhaps more to the point, what would one have to figure out > to "solve" the Frame Problem? "The" frame problem (I put "the" in quotes, since some researchers argue that there are several different problems) is, as I see it, basically a question about what kinds of information must be included in "background knowledge". The "problem" is that, when an action is performed that changes the situation it was performed in, some things change and others don't. In some cases, it's obvious (to us humans) what changes and what doesn't. E.g., if I move a brown desk with a book on it from one side of the room to the other, "obviously" the book has also moved (relative to the room, though not (necessarily) relative to the desk) but the desk is still brown. However, how would a program know that, unless we told it? One possible answer is to only tell it what does change, not what doesn't change (i.e., tell it what's *not* "attached to the frame"). But how about a piece of paper on top of the book? And how about the window that was open? And what about the position of the hands on the clock on the wall? "The frame problem" is sometimes to considered to be a problem because it's not clear how much background information of this kind we need to supply. Some people think we need so much info that the problem in insoluble. Others point out that *we* are not bothered about it, so why should a robot? For more info, see the references and links at: http://www.cse.buffalo.edu/~rapaport/676/F01/sitcalc.html From - Fri Nov 2 08:49:12 2001 Path: acsu.buffalo.edu!pollux.cse.Buffalo.EDU!pvg From: Paul V Gestwicki Newsgroups: sunyab.cse.676 Subject: Re: charting Date: Thu, 1 Nov 2001 15:22:45 -0500 Organization: University at Buffalo Lines: 30 Distribution: sunyab Message-ID: References: <9rak7n$eke$1@prometheus.acsu.buffalo.edu> <3BD961F0.92BD7E5@cse.buffalo.edu> <3BD98B52.95D3941F@cse.buffalo.edu> NNTP-Posting-Host: pollux.cse.buffalo.edu Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII In-Reply-To: <3BD98B52.95D3941F@cse.buffalo.edu> Xref: acsu.buffalo.edu sunyab.cse.676:75 > type 'gimp' in unix or linux - (gimp == photoshop ^ freehand) > you can use it to draw or to convert image file types. > Gimp is a wonderful graphics tool, but a much easier way to convert images between many formats is to use the *nix command 'convert'. It will deduce the transformation required automatically as long as you use standard file extensions (or maybe it uses 'file'... i'm not sure). Example (where % is a *nix prompt) % convert graph1.png graph1.jpg Yes friends, it's that easy! :) -pvg PS - to add to rob's commentary, I suggest Mandrake Linux if you are interested in trying out linux on your home PC. (www.linux-mandrake.com) PPS - And watch out for passport and windows XP. It's all marketing hype and uses unsafe technology. I know /I/ wouldn't want all of my credit card information available to the first guy who cracks the system. http://avirubin.com/passport.html tell your friends From - Mon Nov 5 09:48:42 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: HW #3 QUERY Date: 5 Nov 2001 14:28:12 GMT Organization: University at Buffalo CSE Department Lines: 33 Distribution: sunyab Message-ID: <9s67ls$gcj$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:76 ------------------------------------------------------------------------- Subject: HW #3 QUERY ------------------------------------------------------------------------- A student writes: > I was thinking about the homework again, and am not sure how to > represent the belief report of Anna believes that John bought Stu's dog? > I remember you mentioning several times the difference of de re versus > de dicto. I also noticed a paper you had written in 1987, which I guess > would help me out in figuring out when to use which interpretation. I > also assume from the abstract that in more complicated belief reports > you may want to use both. So I'm not sure how to go about correctly > representing Anna believes that John bought Stu's dog? On the difference between de re and de dicto, see: Rapaport, William J.; Shapiro, Stuart C.; & Wiebe, Janyce M. (1997), "Quasi-Indexicals and Knowledge Reports", Cognitive Science 21: 63-107. http://www.cse.buffalo.edu/~rapaport/Papers/qikrcogsci.pdf (By the way, Jan Wiebe will be the CSE colloquium speaker this Friday!) I would represent the last sentence as de re, which, in this case, means that not both Cassie & Anna believe that the buyer's name is "John", that the seller's name is "Stu", and that the item sold is correctly described as "Stu's dog". I'll discuss this when we go over the homework. In this case, as in most others, there is no "right" answer. From - Mon Nov 12 11:07:11 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: OBJECT/POSSESSOR CASE FRAME Date: 9 Nov 2001 14:22:43 GMT Organization: University at Buffalo CSE Department Lines: 47 Distribution: sunyab Message-ID: <9sgorj$qah$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:78 ------------------------------------------------------------------------- Subject: OBJECT/POSSESSOR CASE FRAME ------------------------------------------------------------------------- A student writes: > I used an object/possessor case frame in my solution to homework #3, and > on my paper, you wrote a comment about my not having a "rel" arec as > part of the case frame. Specifically, I am representing the idea > "Stu's dog" using this case frame. > > In SNePS, would you include a "rel" arc in this case, and if so, what > would you call the relation? It seemed to me that the object/possessor > already encapsulates the idea of ownership. Many of you have asked me a version of this question. Some years ago, Soon Ae Chun, a grad student in both CS and LIN, investigated how to represent general English possessive ("X's Y", or "a Y of X") in SNePS. She decided that there were 3 things that had to be represented: * the possessor * the thing possessed * the kind of thing that was possessed by the possessor and she proposed the "object-possessor-rel" case frame, with the following syntax and semantics: M /|\ / | \ / | \ / | \ possesser/ rel| \object / | \ / | \ v v v p r o [[M]] is the proposition that [[o]] is [[p]]'s [[r]] (OR: ... that [[o]] is an [[r]] of [[p]]) Reference: 1987-5 S. A. Chun. SNePS implementation of possessive phrases. SNeRG Technical Note 19, Department of Computer Science, University at Buffalo, 1987. I will try to put a copy of this on the Web. From - Wed Nov 14 09:15:33 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: CONCEPTUAL DEPENDENCY Date: 13 Nov 2001 19:50:05 GMT Organization: University at Buffalo CSE Department Lines: 8 Distribution: sunyab Message-ID: <9srthd$deh$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:79 ------------------------------------------------------------------------- Subject: CONCEPTUAL DEPENDENCY ------------------------------------------------------------------------- The CD webpage has been updated; it is at: http://www.cse.buffalo.edu/~rapaport/676/F01/cd.html From - Fri Nov 16 09:14:52 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: WOODS ON QUANTIFIERS IN SEMANTIC NETWORKS Date: 16 Nov 2001 14:14:09 GMT Organization: University at Buffalo CSE Department Lines: 52 Distribution: sunyab Message-ID: <9t36vh$mim$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:80 ------------------------------------------------------------------------- Subject: WOODS ON QUANTIFIERS IN SEMANTIC NETWORKS ------------------------------------------------------------------------- Reference: ---------- Woods, William A. (1975), "What's in a Link: Foundations for Semantic Networks", in Brachman, Ronald J. & Levesque, Hector J. (eds.) (1985), Readings in Knowledge Representation (Los Altos, CA: Morgan Kaufmann): 237-238. Woods's representation of: (forall x : Integer(x) ^ T(x))(exists y : T(y))[x is greater than y] in SNePSUL notation: (assert type quant quant-type every variable x class integer restriction T prop (build type quant quant-type some variable y restriction T prop (build type proposition verb greater arg1 x arg2 y))) SNePSUL representation of a modern SNePS representation: (assert forall $x &ant (build member *x class integer) &ant (build object *x property T) cq (build exists $y cq (build object *y property T) cq (build rel greater arg1 *x arg2 *y))) Note that the main difference is that, where Woods represents both the fact that there is a quantifier and the type of quantifier (universal or existential) by nodes, SNePS uses an arc label for the type and doesn't otherwise explicitly show that it is a quantifier. It is SNIP, the SNePS Inference Package, that embodies the "knowledge" that this is a quantified proposition. (Actually, since the current version of SNePS does not implement the existential quantifier, we would need to use Skolem functions.) From - Mon Nov 26 09:37:00 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: NEED COPIES OF C.V.A. PASSAGES Date: 20 Nov 2001 14:17:02 GMT Organization: University at Buffalo CSE Department Lines: 24 Distribution: sunyab Message-ID: <9tdoku$pck$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:81 ------------------------------------------------------------------------- Subject: NEED COPIES OF C.V.A. PASSAGES ------------------------------------------------------------------------- This message only concerns those students doing CVA passages, viz., Marc K Broklawski "proliferate" Chienchih Chi "sedative" Paul V Gestwicki "oam" Kazuhiro Kawachi "taciturn" Valerie Yakich "proximity" Please send me a plain-text file containing the complete passage that you are representing. (You may have sent me this before, but I need to send it to one of the other research-group members, and I want to make sure I have exactly what you've been working on.) I would like this ASAP, please; thanks! -Bill From - Mon Nov 26 09:37:09 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: REVISED SYLLABUS Date: 22 Nov 2001 14:52:47 GMT Organization: University at Buffalo CSE Department Lines: 9 Distribution: sunyab Message-ID: <9tj3fv$83m$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: castor.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:82 ------------------------------------------------------------------------- Subject: REVISED SYLLABUS ------------------------------------------------------------------------- I have made what I expect will be the final revisions to the syllabus, updating the final topics and readings. See: http://www.cse.buffalo.edu/~rapaport/676/F01/syl.html From - Wed Nov 28 09:04:53 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: POSSESSIVE CASE FRAME Date: 27 Nov 2001 14:27:50 GMT Organization: University at Buffalo CSE Department Lines: 11 Distribution: sunyab Message-ID: <9u07t6$81u$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:83 ------------------------------------------------------------------------- Subject: POSSESSIVE CASE FRAME ------------------------------------------------------------------------- Soon Ae Chun's tech report on the object-possessor-rel case frame is now online in the Bison reserve section for our course. It is listed in the index as: SNEPS IMPLEMENTATION OF P(CHUN) From - Wed Nov 28 09:05:02 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: FRAMES Date: 27 Nov 2001 20:13:14 GMT Organization: University at Buffalo CSE Department Lines: 9 Distribution: sunyab Message-ID: <9u0s4q$hfr$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:84 ------------------------------------------------------------------------- Subject: FRAMES ------------------------------------------------------------------------- I've update the webpage on Frames, to include a reference to the paper by Ben Kuipers that I mentioned in lecture. Go to: http://www.cse.buffalo.edu/~rapaport/676/F01/frames.html From - Wed Nov 28 09:05:12 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: SNePS DESIGN QUESTION Date: 27 Nov 2001 20:24:18 GMT Organization: University at Buffalo CSE Department Lines: 17 Distribution: sunyab Message-ID: <9u0spi$hp2$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:85 ------------------------------------------------------------------------- Subject: SNePS DESIGN QUESTION ------------------------------------------------------------------------- A student writes: > When using a shortened form of a case frame, such as agent/act/object > where the object is unknown, is it preferable in SNePS to include the > unused arc and have it point to a base node, or to omit the arc > altogether? For my report, I am currently eliding the unused arc to > make the diagrams more legible, but since arcs can not be added later, > I wasn't sure if it might be considered better SNePS style to include > all arcs of a case frame. Good question. My inclination would be to include the unused arc and have it point to a base node, for 2 reasons: First, there's the reason you give, that that missing info cannot be added later. Second, including it is more in the style of CD and Frames, where you give all "slot" info, but use "default" values when the "fillers" are unknown. From - Wed Nov 28 16:44:00 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: HOW TO FIND THE CHUN ARTICLE ON BISON Date: 28 Nov 2001 16:44:10 GMT Organization: University at Buffalo CSE Department Lines: 25 Distribution: sunyab Message-ID: <9u348q$q8f$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:86 ------------------------------------------------------------------------- Subject: HOW TO FIND THE CHUN ARTICLE ON BISON ------------------------------------------------------------------------- A student writes: | I logged in to ublibraries and did a search for "sneps", but this | article is not listed. Is it not listed yet or did I go about it | incorrectly somehow? Do the following: 1. Log on to Bison 2. Click on Course Reserve (on the bottom, towards the left) 3. Enter "CSE676" (all caps, no spaces) in "Course No." 4. Start Search 5. Click "Next Page" 6. The Chun article is the last one listed Helpful Hint: While you are in the Course Reserve service, do *NOT* use your Web browser's "back" or "forward" buttons. ONLY use the buttons provided by the Course Reserve service to navigate. If you try to use your browser's navigation buttons, weird things will happen to the links (they'll all point to the last item you looked at)! From - Mon Dec 3 09:19:37 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: ANOTHER SNePS QUESTION: SEMANTICS Date: 3 Dec 2001 14:12:19 GMT Organization: University at Buffalo CSE Department Lines: 20 Distribution: sunyab Message-ID: <9ug183$c4e$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:87 ------------------------------------------------------------------------- Subject: ANOTHER SNePS QUESTION: SEMANTICS ------------------------------------------------------------------------- A student writes: > In Friday's presentation, you proposed using a case frame with two > 'obj1' arcs. In a case like this, using an arc label multiple times, > would the case frame semantics be specified differently? For example, > would you describe a "arg1/arg1/rel/arg2" case frame or stick with > "arg1/rel/arg2" and mention in the semantics that there may be more > than one 'arg1'? Or perhaps no change needs to be made? Probably the best thing to do would be to define a "rel/arg1/.../arg1/arg2" case frame, indicating that there are >=1 "arg1" arcs, just as the "official" definition of the "rel/arg1/arg2" case frame is actually for a "rel/arg1/.../argN" case frame, with N >=2 (I think); check out the Case Frame Dictionary. In SNePS 3, this is provided for, since each arc label specifies how many "copies" of it are legal. From - Tue Dec 4 09:32:54 2001 Path: acsu.buffalo.edu!pollux.cse.Buffalo.EDU!pvg From: Paul V Gestwicki Newsgroups: sunyab.cse.676 Subject: SNePS Dia sheet Date: Mon, 3 Dec 2001 11:23:59 -0500 Organization: University at Buffalo Lines: 36 Distribution: sunyab Message-ID: NNTP-Posting-Host: pollux.cse.buffalo.edu Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII NNTP-Posting-User: pvg Xref: acsu.buffalo.edu sunyab.cse.676:88 Greetings In case you're using dia (the tool Rob mentioned earlier in the semester), I have a SNePS shape & sheet you might like to use. I got sick of drawing circles, putting text in them, and then grouping them together, so I put together a simple node shape that does the busy work and sets some connection points. You can get the file at: ~pvg/pub/sneps-dia.tar.gz On *nix, copy this file into your .dia directory and extract it. It will create a sheets folder and a shapes folder. Next time you start up dia, it should have the SNePS sheet ready for you. For those of you who haven't checked it out, here's the dia url again: http://www.lysator.liu.se/~alla/dia/ I create the diagrams in dia, export them as png images, convert them to jpg (a la "convert im.png im.jpg") and then import them into StarOffice 6.0 beta for a beauty of a free-software solution :) Also, I spent last Wednesday checking out 'dot', a graph drawing system developed by AT&T research. Turns out that the algorithm they're using may be modifiable to be used in my own research, but that's beside the point. My conclusion regarding cse676 is that due to dot's leaning towards strict hierarchical drawings, it can not be used to make very pretty SNePS diagrams. The samples I tried all came out with a huge number of edge crossings due to the layering algorithm. The reason I share is so if you say "ooh, maybe I can use 'dot'," perhaps I can save you a few hours of time :) Cheers, Paul From - Wed Dec 5 08:55:01 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: C.V.A. ALGORITHM Date: 4 Dec 2001 18:25:47 GMT Organization: University at Buffalo CSE Department Lines: 16 Distribution: sunyab Message-ID: <9uj4fb$2vo$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:89 ------------------------------------------------------------------------- Subject: C.V.A. ALGORITHM ------------------------------------------------------------------------- Three different versions of Karen Ehrlich's CVA algorithm are now available on the web at: http://www.cse.buffalo.edu/~rapaport/676/F01/cva.html (or see the Directory of Documents). I may have forgotten to tell you that I also posted some good cartoons on the Directory... :-); see: http://www.cse.buffalo.edu/~rapaport/676/F01/directory.html From - Wed Dec 5 08:57:14 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: TERM PROJECT INFORMATION Date: 5 Dec 2001 13:54:07 GMT Organization: University at Buffalo CSE Department Lines: 27 Distribution: sunyab Message-ID: <9ul8tv$5id$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:90 ------------------------------------------------------------------------- Subject: TERM PROJECT INFORMATION ------------------------------------------------------------------------- Please re-read the Knowledge-Representation Guidelines Writing Guidelines at: http://www.cse.buffalo.edu/~rapaport/676/F01/krguidelines.html http://www.cse.buffalo.edu/~rapaport/typing.info.pdf This is especially important with respect to such items as: what needs to be included, required abstract, no cover page, proper formatting, etc. An index to all relevant documents is on the Term Project page at: http://www.cse.buffalo.edu/~rapaport/676/F01/termproj.html From - Wed Dec 5 09:33:15 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: WHEN IS A NODE "BELIEVED" BY THE SYSTEM? Date: 5 Dec 2001 14:29:13 GMT Organization: University at Buffalo CSE Department Lines: 25 Distribution: sunyab Message-ID: <9ulavp$6fh$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:91 ------------------------------------------------------------------------- Subject: WHEN IS A NODE "BELIEVED" BY THE SYSTEM? ------------------------------------------------------------------------- A student asks: > The "SNePS Considered as a Fully Intensional Propositional Network" > article states that structured proposition nodes are > believed by the system if they have no incoming arcs. In my first > sentence of my passage, which has the form "x shows that y has > property F", I've represented "y has property F" on the highest level > with asserted proposition node (consisiting of a Object-Property case > frame). I point to this later on when constructing part indicating that > x itself shows this. Can this really be a belief of the system since I > have an incoming arc pointing to it? That paper refers to an older version of SNePS. In the current version, any *asserted* node, at any level, with or without incoming arcs, is "believed" by the system. Nodes that are *not* asserted are either proposition nodes that the system does not believe (which is not the same as saying that the system believes that it is not the case!)--i.e., it is just "considering" them--or nodes that are not proposition nodes, e.g., nodes representing concepts or individuals. From - Fri Dec 7 09:07:07 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: ALMEIDA ON REPRESENTING TIME IN SNePS Date: 6 Dec 2001 18:33:08 GMT Organization: University at Buffalo CSE Department Lines: 22 Distribution: sunyab Message-ID: <9uodl4$14q$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:92 ------------------------------------------------------------------------- Subject: ALMEIDA ON REPRESENTING TIME IN SNePS ------------------------------------------------------------------------- The following paper: Michael J. Almeida. Time in narratives. In Judith F. Duchan, Gail A. Bruder, and Lynne E. Hewitt, editors, Deixis in Narrative: A Cognitive Science Perspective, pages 159-189. Lawrence Erlbaum Associates, Inc., Hillsdale, NJ, 1995. is now online at either: http://ublib.buffalo.edu/cgi-bin/reserve.cgi?B027575624.PDF or at: http://www.cse.buffalo.edu/~rapaport/676/F01/almeida95.pdf From - Fri Dec 7 14:02:33 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: TERM PROJECTS Date: 7 Dec 2001 18:46:29 GMT Organization: University at Buffalo CSE Department Lines: 23 Distribution: sunyab Message-ID: <9ur2q5$ene$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: castor.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:93 ------------------------------------------------------------------------- Subject: TERM PROJECTS ------------------------------------------------------------------------- Your term projects are due Monday, Dec. 10. If you need extra time, please let me know by email. Please turn in your paper either by giving it to me in person in my office, Bell 214 OR putting it in my mailbox in Bell 211 OR coming back the next day to do one of the above; i.e., please do NOT slip them under my door, where they are sure to get lost :-) I will post news when they are ready to be picked up. From - Mon Dec 10 09:43:43 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: MOLECULAR VS. PATTERN NODES Date: 10 Dec 2001 14:31:46 GMT Organization: University at Buffalo CSE Department Lines: 21 Distribution: sunyab Message-ID: <9v2h0i$bvp$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:94 ------------------------------------------------------------------------- Subject: MOLECULAR VS. PATTERN NODES ------------------------------------------------------------------------- A student writes: > In my diagrams, I have my pattern nodes labeled as Mn (where n is some > integer) where it appears they should be Pn. Is this purely a matter > of formality, or is it incorrect to label them Mn? I am assuming that > it is a case where, if input to SNePS, it would make up its own nodes > and call them Pn anyway, so it shouldn't matter which way I go in my > diagrams. It is technically incorrect (though in previous versions of SNePS, it was OK) to label a molecular node that dominates a variable node with "Mn"; it is, technically, a "pattern" node, to be labeled "Pn". But you are correct that when inputting this in SNePSUL, SNePS makes the decision for you. I won't mark any of you "wrong" if you make this error (which is on a par with omitting a semicolon in a program whose compiler adds missing semicolons for you automatically). From - Mon Dec 10 09:43:50 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: HOW MUCH BACKGROUND KNOWLEDGE SHOULD BE REPRESENTED? Date: 10 Dec 2001 14:40:37 GMT Organization: University at Buffalo CSE Department Lines: 12 Distribution: sunyab Message-ID: <9v2hh5$c6u$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:95 ------------------------------------------------------------------------- Subject: HOW MUCH BACKGROUND KNOWLEDGE SHOULD BE REPRESENTED? ------------------------------------------------------------------------- Some of you have asked me how much background knowledge needs to be represented. According to the grading criteria that were posted in October and what I've been saying all along, you need to represent *some* background knowledge. However, it will suffice if at least you specify what background knowledge is needed, in English even if not coded in your KR language. From - Tue Dec 18 15:08:53 2001 Path: acsu.buffalo.edu!rapaport From: rapaport@cse.buffalo.edu (William J. Rapaport) Newsgroups: sunyab.cse.676 Subject: FINAL GRADES Date: 14 Dec 2001 18:20:46 GMT Organization: University at Buffalo CSE Department Lines: 38 Distribution: sunyab Message-ID: <9vdftu$so0$1@prometheus.acsu.buffalo.edu> NNTP-Posting-Host: adara.cse.buffalo.edu NNTP-Posting-User: rapaport Xref: acsu.buffalo.edu sunyab.cse.676:96 ------------------------------------------------------------------------- Subject: FINAL GRADES ------------------------------------------------------------------------- I have finished grading your term projects and computing your course grades. If you are in CSE, I will put your term project in your mailbox or mail folder. Otherwise, you can stop by any time I'm in my office to pick it up, either now (at least till next Tuesday) or next semester (contact me first to make sure I'll be in); or else you can give me a campus or postal mailing address to send it to. I was generally pleased with your projects, and I'm sure they will be useful as we continue work on the CVA and HDS projects. So, thanks! And now... (drum roll) ... the grades: ========================================================================= CSE 676 KNOWLEDGE REPRESENTATION Fall 2001 version: grades.14dc01.3 ------------------------------------------------------------------------- Person Term Course Number Paper Grade [CG := (TP + 2*"A")/3, where "A" = 4.0] 21306247 B+ A- 24964928 B+ A- 26415173 B+ A- 26893138 A A 26901293 A A 27572746 A A 27697539 A- A 27865612 A- A 29031204 A A