From owner-cse727-sp07-list@LISTSERV.BUFFALO.EDU Sun Apr 8 15:47:53 2007 Received: from ares.cse.buffalo.edu (ares.cse.Buffalo.EDU [128.205.32.79]) by castor.cse.Buffalo.EDU (8.13.6/8.12.10) with ESMTP id l38JlrCo001853 for ; Sun, 8 Apr 2007 15:47:53 -0400 (EDT) Received: from front1.acsu.buffalo.edu (coldfront.acsu.buffalo.edu [128.205.6.89]) by ares.cse.buffalo.edu (8.13.6/8.13.6) with SMTP id l38Jlp1k084024 for ; Sun, 8 Apr 2007 15:47:51 -0400 (EDT) Received: (qmail 19813 invoked from network); 8 Apr 2007 19:47:51 -0000 Received: from mailscan5.acsu.buffalo.edu (128.205.6.137) by front1.acsu.buffalo.edu with SMTP; 8 Apr 2007 19:47:51 -0000 Received: (qmail 26037 invoked from network); 8 Apr 2007 19:47:50 -0000 Received: from deliverance.acsu.buffalo.edu (128.205.7.57) by front2.acsu.buffalo.edu with SMTP; 8 Apr 2007 19:47:50 -0000 Received: (qmail 12857 invoked from network); 8 Apr 2007 19:47:36 -0000 Received: from listserv.buffalo.edu (128.205.7.35) by deliverance.acsu.buffalo.edu with SMTP; 8 Apr 2007 19:47:36 -0000 Received: by LISTSERV.BUFFALO.EDU (LISTSERV-TCP/IP release 14.5) with spool id 4461525 for CSE727-SP07-LIST@LISTSERV.BUFFALO.EDU; Sun, 8 Apr 2007 15:47:36 -0400 Delivered-To: cse727-sp07-list@listserv.buffalo.edu Received: (qmail 9091 invoked from network); 8 Apr 2007 19:47:36 -0000 Received: from mailscan1.acsu.buffalo.edu (128.205.6.133) by listserv.buffalo.edu with SMTP; 8 Apr 2007 19:47:36 -0000 Received: (qmail 27818 invoked by uid 60001); 8 Apr 2007 19:47:35 -0000 X-Mailer: University at Buffalo WebMail Cyrusoft SilkyMail v1.1.11 MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 8bit X-Originating-IP: 66.67.170.95 X-UB-Relay: (internal) X-PM-EL-Spam-Prob: : 7% Message-ID: <1176061655.461946d7e7714@mail1.buffalo.edu> Date: Sun, 8 Apr 2007 15:47:35 -0400 Reply-To: jcwhite2@BUFFALO.EDU Sender: "CVA Seminar, Spring 2007" From: John C White Subject: CSE727: Responce to hastingslytinen94a To: CSE727-SP07-LIST@LISTSERV.BUFFALO.EDU Precedence: list List-Help: , List-Unsubscribe: List-Subscribe: List-Owner: List-Archive: X-UB-Relay: (deliverance.acsu.buffalo.edu) X-DCC-Buffalo.EDU-Metrics: castor.cse.Buffalo.EDU 1336; Body=0 Fuz1=0 Fuz2=0 X-Spam-Status: No, score=-2.5 required=5.0 tests=AWL,BAYES_00 autolearn=unavailable version=3.1.7 X-Spam-Checker-Version: SpamAssassin 3.1.7 (2006-10-05) on ares.cse.buffalo.edu X-Virus-Scanned: ClamAV 0.88.6/3044/Sun Apr 8 12:50:42 2007 on ares.cse.buffalo.edu X-Virus-Status: Clean Status: R Content-Length: 969 After reading "The Ups and Downs of Lexical Acquisition," I have some idea how the system they put together by the authors, Camille, is all about. Though lacking details about the exact implementation of Camille, the fact that it builds lexical networks then performs a graphical search on them in order to locate the node(s) suggests a similar method of acquisition to Cassie. I find it interesting how it narrows down the definition of verbs to be more specific while making noun definitions more general. Does this really correspond to how a human acquires vocabulary? Though that may not necessarily be the goal of the authors, I would think the most human approach to CVA would be similar for both nouns and verbs in this respect. Why does it make more sense to narrow down the verb but expand the possible meanings of the nouns? Is it just the case with this system that such a method was chosen in order to get the least incorrect results? -John White From owner-cse727-sp07-list@LISTSERV.BUFFALO.EDU Sun Apr 8 23:07:41 2007 Received: from ares.cse.buffalo.edu (ares.cse.Buffalo.EDU [128.205.32.79]) by castor.cse.Buffalo.EDU (8.13.6/8.12.10) with ESMTP id l3937fiT000030 for ; Sun, 8 Apr 2007 23:07:41 -0400 (EDT) Received: from front2.acsu.buffalo.edu (upfront.acsu.buffalo.edu [128.205.4.140]) by ares.cse.buffalo.edu (8.13.6/8.13.6) with SMTP id l3937dUc011191 for ; Sun, 8 Apr 2007 23:07:39 -0400 (EDT) Received: (qmail 22859 invoked from network); 9 Apr 2007 03:07:39 -0000 Received: from mailscan7.acsu.buffalo.edu (128.205.6.158) by front2.acsu.buffalo.edu with SMTP; 9 Apr 2007 03:07:39 -0000 Received: (qmail 22849 invoked from network); 9 Apr 2007 03:07:39 -0000 Received: from deliverance.acsu.buffalo.edu (128.205.7.57) by front2.acsu.buffalo.edu with SMTP; 9 Apr 2007 03:07:39 -0000 Received: (qmail 21934 invoked from network); 9 Apr 2007 03:07:35 -0000 Received: from listserv.buffalo.edu (128.205.7.35) by deliverance.acsu.buffalo.edu with SMTP; 9 Apr 2007 03:07:35 -0000 Received: by LISTSERV.BUFFALO.EDU (LISTSERV-TCP/IP release 14.5) with spool id 4467402 for CSE727-SP07-LIST@LISTSERV.BUFFALO.EDU; Sun, 8 Apr 2007 23:07:35 -0400 Delivered-To: CSE727-SP07-LIST@listserv.buffalo.edu Received: (qmail 19683 invoked from network); 9 Apr 2007 02:57:35 -0000 Received: from mailscan8.acsu.buffalo.edu (128.205.7.55) by listserv.buffalo.edu with SMTP; 9 Apr 2007 02:57:35 -0000 Received: (qmail 21717 invoked from network); 9 Apr 2007 02:57:34 -0000 Received: from nz-out-0506.google.com (64.233.162.236) by smtp4.acsu.buffalo.edu with SMTP; 9 Apr 2007 02:57:34 -0000 Received: by nz-out-0506.google.com with SMTP id z31so667526nzd for ; Sun, 08 Apr 2007 19:57:34 -0700 (PDT) Received: by 10.115.46.9 with SMTP id y9mr2136631waj.1176087454128; Sun, 08 Apr 2007 19:57:34 -0700 (PDT) Received: by 10.115.73.2 with HTTP; Sun, 8 Apr 2007 19:57:34 -0700 (PDT) MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Content-Disposition: inline X-Google-Sender-Auth: 0b2df64a1dd1f318 X-UB-Relay: (nz-out-0506.google.com) X-PM-Spam-Prob: : 7% Message-ID: <4e4445330704081957x2699ba55s5b2a73959cd14a@mail.gmail.com> Date: Sun, 8 Apr 2007 22:57:34 -0400 Reply-To: "Paul M. Heider" Sender: "CVA Seminar, Spring 2007" From: "Paul M. Heider" Subject: CSE727: hastings & lytinen 1994b "Objects, actions, nouns, and verbs" To: CSE727-SP07-LIST@LISTSERV.BUFFALO.EDU Precedence: list List-Help: , List-Unsubscribe: List-Subscribe: List-Owner: List-Archive: X-UB-Relay: (nz-out-0506.google.com) X-DCC-Buffalo.EDU-Metrics: castor.cse.Buffalo.EDU 1029; Body=0 Fuz1=0 Fuz2=0 X-Spam-Status: No, score=-2.5 required=5.0 tests=AWL,BAYES_00 autolearn=unavailable version=3.1.7 X-Spam-Checker-Version: SpamAssassin 3.1.7 (2006-10-05) on ares.cse.buffalo.edu X-Virus-Scanned: ClamAV 0.88.6/3050/Sun Apr 8 19:47:56 2007 on ares.cse.buffalo.edu X-Virus-Status: Clean Status: R Content-Length: 815 It seems from the description that all matches have to occur with a prior existing word. That is, they are matching unknown words to known words in the tree. Berwick's algorithm needs only to return the N closest matches and that's your relative definition. I don't know how to modify H&L's algorithm to actual create new concepts. What if we used some sort of anti-collision technique to create unique places in the hierarchy. That is, since H&L don't use negative examples, we could treat all the other sentences with a similar environment to be a negative example. This also opens up cans about who gets the more specific definition, what to do when we don't have enough definition to create a judgement (maybe we just force a judgement? hash to the n+1 position?), etc... I think it has potential. Paul From owner-cse727-sp07-list@LISTSERV.BUFFALO.EDU Sun Apr 8 23:08:27 2007 Received: from ares.cse.buffalo.edu (ares.cse.Buffalo.EDU [128.205.32.79]) by castor.cse.Buffalo.EDU (8.13.6/8.12.10) with ESMTP id l3938Qjr000102 for ; Sun, 8 Apr 2007 23:08:27 -0400 (EDT) Received: from front3.acsu.buffalo.edu (warmfront.acsu.buffalo.edu [128.205.6.88]) by ares.cse.buffalo.edu (8.13.6/8.13.6) with SMTP id l3938OtS011237 for ; Sun, 8 Apr 2007 23:08:24 -0400 (EDT) Received: (qmail 9391 invoked from network); 9 Apr 2007 03:08:24 -0000 Received: from mailscan8.acsu.buffalo.edu (128.205.7.55) by front3.acsu.buffalo.edu with SMTP; 9 Apr 2007 03:08:24 -0000 Received: (qmail 13638 invoked from network); 9 Apr 2007 03:08:24 -0000 Received: from deliverance.acsu.buffalo.edu (128.205.7.57) by front1.acsu.buffalo.edu with SMTP; 9 Apr 2007 03:08:24 -0000 Received: (qmail 22707 invoked from network); 9 Apr 2007 03:08:20 -0000 Received: from listserv.buffalo.edu (128.205.7.35) by deliverance.acsu.buffalo.edu with SMTP; 9 Apr 2007 03:08:20 -0000 Received: by LISTSERV.BUFFALO.EDU (LISTSERV-TCP/IP release 14.5) with spool id 4467427 for CSE727-SP07-LIST@LISTSERV.BUFFALO.EDU; Sun, 8 Apr 2007 23:08:20 -0400 Delivered-To: CSE727-SP07-LIST@listserv.buffalo.edu Received: (qmail 20018 invoked from network); 9 Apr 2007 02:58:20 -0000 Received: from mailscan5.acsu.buffalo.edu (128.205.6.137) by listserv.buffalo.edu with SMTP; 9 Apr 2007 02:58:20 -0000 Received: (qmail 21301 invoked from network); 9 Apr 2007 02:58:20 -0000 Received: from wr-out-0506.google.com (64.233.184.225) by smtp3.acsu.buffalo.edu with SMTP; 9 Apr 2007 02:58:20 -0000 Received: by wr-out-0506.google.com with SMTP id 37so836568wra for ; Sun, 08 Apr 2007 19:58:19 -0700 (PDT) Received: by 10.114.25.3 with SMTP id 3mr2143234way.1176087499162; Sun, 08 Apr 2007 19:58:19 -0700 (PDT) Received: by 10.115.73.2 with HTTP; Sun, 8 Apr 2007 19:58:19 -0700 (PDT) MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Content-Disposition: inline X-Google-Sender-Auth: 647470e5b8efc5ac X-UB-Relay: (wr-out-0506.google.com) X-PM-Spam-Prob: : 7% Message-ID: <4e4445330704081958m6bc0b5t297a02d33de3281@mail.gmail.com> Date: Sun, 8 Apr 2007 22:58:19 -0400 Reply-To: "Paul M. Heider" Sender: "CVA Seminar, Spring 2007" From: "Paul M. Heider" Subject: CSE727: Hastings & Lytinen 1994a The Ups and Downs of Lexical Acquisition To: CSE727-SP07-LIST@LISTSERV.BUFFALO.EDU Precedence: list List-Help: , List-Unsubscribe: List-Subscribe: List-Owner: List-Archive: X-UB-Relay: (wr-out-0506.google.com) X-DCC-Buffalo.EDU-Metrics: castor.cse.Buffalo.EDU 1029; Body=0 Fuz1=0 Fuz2=0 X-Spam-Status: No, score=-2.5 required=5.0 tests=AWL,BAYES_00 autolearn=ham version=3.1.7 X-Spam-Checker-Version: SpamAssassin 3.1.7 (2006-10-05) on ares.cse.buffalo.edu X-Virus-Scanned: ClamAV 0.88.6/3050/Sun Apr 8 19:47:56 2007 on ares.cse.buffalo.edu X-Virus-Status: Clean Status: R Content-Length: 1217 H&L got halfway through answering two of my questions from their 1994b article (which I read first, by accident). I guess I'll have to read Lytinen 1988 (footnote on page 756) to get a full answer on vague vs. polysemous versus ambiguous. The context I was puzzling over is a word that (so far) tightly fits two "cousin" terms. Let's say it is a noun that seems to match "giraffe" and "rhinoceros." If you walk up the animal tree far enough, they will connect. I was curious when it would be best to choose "animal" as the _actual_ meaning and when to give up and consider this word to mean either "giraffe" or "rhinoceros." The second puzzle from 1994b was only half solved because, while 1994a includes a complete answer, they also make an assumption that I find very untrue to the real world. It seems they have objectively separated their "concepts" from their "words." Of course, there is a one-to-one correlation between the two (except when they have removed a "word" to let it be assigned to a "concept"). Exact matches to atomic elements is a little deterministic as to what your machine is allowed to learn. It removes alot of the "uncoded" aspect of the algorithm and removes scalability. ~Paul From owner-cse727-sp07-list@LISTSERV.BUFFALO.EDU Fri Apr 13 04:37:27 2007 Received: from ares.cse.buffalo.edu (ares.cse.Buffalo.EDU [128.205.32.79]) by castor.cse.Buffalo.EDU (8.13.6/8.12.10) with ESMTP id l3D8bRXj027828 for ; Fri, 13 Apr 2007 04:37:27 -0400 (EDT) Received: from front3.acsu.buffalo.edu (warmfront.acsu.buffalo.edu [128.205.6.88]) by ares.cse.buffalo.edu (8.13.6/8.13.6) with SMTP id l3D8bOTA046362 for ; Fri, 13 Apr 2007 04:37:24 -0400 (EDT) Received: (qmail 17176 invoked from network); 13 Apr 2007 08:37:24 -0000 Received: from mailscan1.acsu.buffalo.edu (128.205.6.133) by front3.acsu.buffalo.edu with SMTP; 13 Apr 2007 08:37:24 -0000 Received: (qmail 17168 invoked from network); 13 Apr 2007 08:37:23 -0000 Received: from deliverance.acsu.buffalo.edu (128.205.7.57) by front3.acsu.buffalo.edu with SMTP; 13 Apr 2007 08:37:23 -0000 Received: (qmail 5843 invoked from network); 13 Apr 2007 08:37:21 -0000 Received: from listserv.buffalo.edu (128.205.7.35) by deliverance.acsu.buffalo.edu with SMTP; 13 Apr 2007 08:37:21 -0000 Received: by LISTSERV.BUFFALO.EDU (LISTSERV-TCP/IP release 14.5) with spool id 4681821 for CSE727-SP07-LIST@LISTSERV.BUFFALO.EDU; Fri, 13 Apr 2007 04:37:21 -0400 Delivered-To: CSE727-SP07-LIST@LISTSERV.BUFFALO.EDU Received: (qmail 8825 invoked from network); 13 Apr 2007 08:37:21 -0000 Received: from mailscan3.acsu.buffalo.edu (128.205.6.135) by listserv.buffalo.edu with SMTP; 13 Apr 2007 08:37:21 -0000 Received: (qmail 29083 invoked from network); 13 Apr 2007 08:37:20 -0000 Received: from 67-20-233-65.kntnny.adelphia.net (HELO pavano) (67.20.233.65) by smtp4.acsu.buffalo.edu with SMTP; 13 Apr 2007 08:37:20 -0000 MIME-Version: 1.0 Content-Type: text/plain; format=flowed; charset="iso-8859-1"; reply-type=original Content-Transfer-Encoding: 7bit X-Priority: 3 X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook Express 6.00.2900.3028 X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.3028 X-UB-Relay: (67-20-233-65.kntnny.adelphia.net) X-PM-Spam-Prob: XX: 22% Message-ID: <001b01c77da6$f299d740$41e91443@pavano> Date: Fri, 13 Apr 2007 04:37:18 -0400 Reply-To: "J. A." Sender: "CVA Seminar, Spring 2007" From: "J. A." Subject: CSE727: Comments on Hastings & Lytinen 1994 To: CSE727-SP07-LIST@LISTSERV.BUFFALO.EDU Precedence: list List-Help: , List-Unsubscribe: List-Subscribe: List-Owner: List-Archive: X-UB-Relay: (67-20-233-65.kntnny.adelphia.net) X-DCC-Buffalo.EDU-Metrics: castor.cse.Buffalo.EDU 1335; Body=0 Fuz1=0 Fuz2=0 X-Spam-Status: No, score=-2.1 required=5.0 tests=AWL,BAYES_00 autolearn=ham version=3.1.7 X-Spam-Checker-Version: SpamAssassin 3.1.7 (2006-10-05) on ares.cse.buffalo.edu X-Virus-Scanned: ClamAV 0.88.6/3088/Thu Apr 12 22:05:26 2007 on ares.cse.buffalo.edu X-Virus-Status: Clean Status: R Content-Length: 1186 The program demonstrated in this paper seems to get an "okay" approximation of an unknown word by using a fairly rich set of domain knowledge. Just as in CVA, the algorithm converges to better and better definitions as more and more examples of the unknown word (in context) are given to it. While I was reading, I wondered what the effects of combining the CVA algorithms with such detailed sets of domain knowledge (such as the military domain given in the paper) would be. As far as NLP is concerned, the CVA techniques could speed up the "process of elimination" for verbs and the reduction of generality for nouns. On the CVA side of things, one of the proposed seminar projects was to integrate CYC with SNePS to provide a very large set of background knowledge. This paper seems to provide good justification for doing so. By augmenting the background knowledge as such and crafting sentence representations to take advantage of it, the resulting definition would probably be very detailed. It doesn't seem like usage of case frames would present a problem either - after all, the set of domain knowledge in the paper uses the is-a relation exclusively. - Joe From owner-cse727-sp07-list@LISTSERV.BUFFALO.EDU Sun Apr 15 23:00:31 2007 Received: from ares.cse.buffalo.edu (ares.cse.Buffalo.EDU [128.205.32.79]) by castor.cse.Buffalo.EDU (8.13.6/8.12.10) with ESMTP id l3G30VOu017962 for ; Sun, 15 Apr 2007 23:00:31 -0400 (EDT) Received: from front2.acsu.buffalo.edu (upfront.acsu.buffalo.edu [128.205.4.140]) by ares.cse.buffalo.edu (8.13.6/8.13.6) with SMTP id l3G2uVuN099868 for ; Sun, 15 Apr 2007 22:56:31 -0400 (EDT) Received: (qmail 21287 invoked from network); 16 Apr 2007 02:56:31 -0000 Received: from mailscan3.acsu.buffalo.edu (128.205.6.135) by front2.acsu.buffalo.edu with SMTP; 16 Apr 2007 02:56:31 -0000 Received: (qmail 4746 invoked from network); 16 Apr 2007 02:56:30 -0000 Received: from deliverance.acsu.buffalo.edu (128.205.7.57) by front3.acsu.buffalo.edu with SMTP; 16 Apr 2007 02:56:30 -0000 Received: (qmail 20268 invoked from network); 16 Apr 2007 02:56:24 -0000 Received: from listserv.buffalo.edu (128.205.7.35) by deliverance.acsu.buffalo.edu with SMTP; 16 Apr 2007 02:56:24 -0000 Received: by LISTSERV.BUFFALO.EDU (LISTSERV-TCP/IP release 14.5) with spool id 4719549 for CSE727-SP07-LIST@LISTSERV.BUFFALO.EDU; Sun, 15 Apr 2007 22:56:24 -0400 Delivered-To: CSE727-SP07-LIST@LISTSERV.BUFFALO.EDU Received: (qmail 27368 invoked from network); 16 Apr 2007 02:56:24 -0000 Received: from mailscan3.acsu.buffalo.edu (128.205.6.135) by listserv.buffalo.edu with SMTP; 16 Apr 2007 02:56:24 -0000 Received: (qmail 7696 invoked from network); 16 Apr 2007 02:56:23 -0000 Received: from email4.acsu.buffalo.edu (HELO email5.acsu.buffalo.edu) (128.205.7.94) by smtp3.acsu.buffalo.edu with SMTP; 16 Apr 2007 02:56:23 -0000 Content-Disposition: inline Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 X-Origin: 72.88.113.50 X-Mailer: AtMail 4.61 - 72.88.113.50 - petrova3@buffalo.edu X-UB-Relay: (email4.acsu.buffalo.edu) X-PM-EL-Spam-Prob: : 7% Message-ID: <3689.1176692183@buffalo.edu> Date: Sun, 15 Apr 2007 22:56:23 -0400 Reply-To: petrova3@buffalo.edu Sender: "CVA Seminar, Spring 2007" From: Nyurguyana Petrova Subject: Peter Hastings, Steven Lytinen (1994) "The Ups and Downs of Lexical Acquisition": To: CSE727-SP07-LIST@LISTSERV.BUFFALO.EDU Precedence: list List-Help: , List-Unsubscribe: List-Subscribe: List-Owner: List-Archive: X-UB-Relay: (email4.acsu.buffalo.edu) X-DCC-Buffalo.EDU-Metrics: castor.cse.Buffalo.EDU 1029; Body=0 Fuz1=0 Fuz2=0 X-Virus-Scanned: ClamAV 0.88.6/3096/Sun Apr 15 18:26:48 2007 on ares.cse.buffalo.edu X-Virus-Status: Clean Content-Transfer-Encoding: 8bit X-MIME-Autoconverted: from quoted-printable to 8bit by castor.cse.Buffalo.EDU id l3G30VOu017962 Status: RO X-Status: X-Keywords: NotJunk JunkRecorded $NotJunk X-UID: 14832 Content-Length: 782 Peter Hastings, Steven Lytinen (1994) "The Ups and Downs of Lexical Acquisition": Although there is still a lot of work to be done on the Camille's project (as the not-so-high accuracy results show), the fact that Camille does not require an additional human guidance is itself really astounding. In "Future work" section (p. 758) the authors point out that Camille does not have an ability "to make inferences based on sequences of actions, the system has no knowledge of the results of actions, their causes, or what goals they might achieve". And they also mention that the addition of such knowledge would significantly enhance Camille's knowledge, but they also point out that it requires an additional resource. What kind of additional resource they are referring to? -Yana From owner-cse727-sp07-list@LISTSERV.BUFFALO.EDU Tue Apr 17 08:48:49 2007 Received: from ares.cse.buffalo.edu (ares.cse.Buffalo.EDU [128.205.32.79]) by castor.cse.Buffalo.EDU (8.13.6/8.12.10) with ESMTP id l3HCmngN021729 for ; Tue, 17 Apr 2007 08:48:49 -0400 (EDT) Received: from front2.acsu.buffalo.edu (upfront.acsu.buffalo.edu [128.205.4.140]) by ares.cse.buffalo.edu (8.13.6/8.13.6) with SMTP id l3HCmgGW027291 for ; Tue, 17 Apr 2007 08:48:42 -0400 (EDT) Received: (qmail 9109 invoked from network); 17 Apr 2007 12:48:42 -0000 Received: from mailscan7.acsu.buffalo.edu (128.205.6.158) by front2.acsu.buffalo.edu with SMTP; 17 Apr 2007 12:48:42 -0000 Received: (qmail 13475 invoked from network); 17 Apr 2007 12:48:42 -0000 Received: from defer.acsu.buffalo.edu (128.205.7.58) by front3.acsu.buffalo.edu with SMTP; 17 Apr 2007 12:48:42 -0000 Received: (qmail 11015 invoked from network); 17 Apr 2007 12:48:27 -0000 Received: from listserv.buffalo.edu (128.205.7.35) by defer.acsu.buffalo.edu with SMTP; 17 Apr 2007 12:48:27 -0000 Received: by LISTSERV.BUFFALO.EDU (LISTSERV-TCP/IP release 14.5) with spool id 4767797 for CSE727-SP07-LIST@LISTSERV.BUFFALO.EDU; Tue, 17 Apr 2007 08:48:27 -0400 Delivered-To: CSE727-SP07-LIST@LISTSERV.BUFFALO.EDU Received: (qmail 23518 invoked from network); 17 Apr 2007 12:48:14 -0000 Received: from mailscan5.acsu.buffalo.edu (128.205.6.137) by listserv.buffalo.edu with SMTP; 17 Apr 2007 12:48:14 -0000 Received: (qmail 14688 invoked from network); 17 Apr 2007 12:48:14 -0000 Received: from castor.cse.buffalo.edu (128.205.32.14) by smtp5.acsu.buffalo.edu with SMTP; 17 Apr 2007 12:48:14 -0000 Received: from castor.cse.Buffalo.EDU (rapaport@localhost [127.0.0.1]) by castor.cse.Buffalo.EDU (8.13.6/8.12.10) with ESMTP id l3HCmDUG021695 for ; Tue, 17 Apr 2007 08:48:13 -0400 (EDT) Received: (from rapaport@localhost) by castor.cse.Buffalo.EDU (8.13.6/8.12.9/Submit) id l3HCmDt2021694 for CSE727-SP07-LIST@LISTSERV.BUFFALO.EDU; Tue, 17 Apr 2007 08:48:13 -0400 (EDT) X-UB-Relay: (castor.cse.buffalo.edu) X-PM-EL-Spam-Prob: : 7% Message-ID: <200704171248.l3HCmDt2021694@castor.cse.Buffalo.EDU> Date: Tue, 17 Apr 2007 08:48:13 -0400 Reply-To: "William J. Rapaport" Sender: "CVA Seminar, Spring 2007" From: "William J. Rapaport" Subject: Hastings & Lytinen (1994) "The Ups and Downs of Lexical Acquisition": To: CSE727-SP07-LIST@LISTSERV.BUFFALO.EDU Precedence: list List-Help: , List-Unsubscribe: List-Subscribe: List-Owner: List-Archive: X-UB-Relay: (castor.cse.buffalo.edu) X-DCC-Buffalo.EDU-Metrics: castor.cse.Buffalo.EDU 1029; Body=0 Fuz1=0 Fuz2=0 X-Spam-Status: No, score=-2.5 required=5.0 tests=AWL,BAYES_00 autolearn=unavailable version=3.1.7 X-Spam-Checker-Version: SpamAssassin 3.1.7 (2006-10-05) on ares.cse.buffalo.edu X-Virus-Scanned: ClamAV 0.88.6/3110/Tue Apr 17 07:57:27 2007 on ares.cse.buffalo.edu X-Virus-Status: Clean Status: R Content-Length: 848 Yana asked: | ... In "Future work" section (p. 758) the authors point out that | Camille does not have an ability "to make inferences based on sequences of | actions, the system has no knowledge of the results of actions, their causes, or | what goals they might achieve". And they also mention that the addition of such | knowledge would significantly enhance Camille's knowledge, but they also point | out that it requires an additional resource. What kind of additional resource | they are referring to? I think they're merely pointing out that if you are going to have the ability to understand actions and their consequences, you'll need a module to handle that that isn't part of the current implementation of Camille. Note that, in principle at least, SNePS has such a module, namely, SNeRE, though we don't currently use it in that way. From owner-cse727-sp07-list@LISTSERV.BUFFALO.EDU Tue Apr 17 08:58:03 2007 Received: from ares.cse.buffalo.edu (ares.cse.Buffalo.EDU [128.205.32.79]) by castor.cse.Buffalo.EDU (8.13.6/8.12.10) with ESMTP id l3HCw3Hx022363 for ; Tue, 17 Apr 2007 08:58:03 -0400 (EDT) Received: from front3.acsu.buffalo.edu (warmfront.acsu.buffalo.edu [128.205.6.88]) by ares.cse.buffalo.edu (8.13.6/8.13.6) with SMTP id l3HCvseN027975 for ; Tue, 17 Apr 2007 08:57:54 -0400 (EDT) Received: (qmail 17276 invoked from network); 17 Apr 2007 12:57:54 -0000 Received: from mailscan6.acsu.buffalo.edu (128.205.7.95) by front3.acsu.buffalo.edu with SMTP; 17 Apr 2007 12:57:54 -0000 Received: (qmail 20852 invoked from network); 17 Apr 2007 12:57:53 -0000 Received: from deliverance.acsu.buffalo.edu (128.205.7.57) by front2.acsu.buffalo.edu with SMTP; 17 Apr 2007 12:57:53 -0000 Received: (qmail 8127 invoked from network); 17 Apr 2007 12:57:40 -0000 Received: from listserv.buffalo.edu (128.205.7.35) by deliverance.acsu.buffalo.edu with SMTP; 17 Apr 2007 12:57:40 -0000 Received: by LISTSERV.BUFFALO.EDU (LISTSERV-TCP/IP release 14.5) with spool id 4768060 for CSE727-SP07-LIST@LISTSERV.BUFFALO.EDU; Tue, 17 Apr 2007 08:57:40 -0400 Delivered-To: CSE727-SP07-LIST@LISTSERV.BUFFALO.EDU Received: (qmail 27402 invoked from network); 17 Apr 2007 12:57:39 -0000 Received: from mailscan4.acsu.buffalo.edu (128.205.6.136) by listserv.buffalo.edu with SMTP; 17 Apr 2007 12:57:39 -0000 Received: (qmail 17357 invoked from network); 17 Apr 2007 12:57:38 -0000 Received: from castor.cse.buffalo.edu (128.205.32.14) by smtp5.acsu.buffalo.edu with SMTP; 17 Apr 2007 12:57:38 -0000 Received: from castor.cse.Buffalo.EDU (rapaport@localhost [127.0.0.1]) by castor.cse.Buffalo.EDU (8.13.6/8.12.10) with ESMTP id l3HCvcLr022349 for ; Tue, 17 Apr 2007 08:57:38 -0400 (EDT) Received: (from rapaport@localhost) by castor.cse.Buffalo.EDU (8.13.6/8.12.9/Submit) id l3HCvc7l022348 for CSE727-SP07-LIST@LISTSERV.BUFFALO.EDU; Tue, 17 Apr 2007 08:57:38 -0400 (EDT) X-UB-Relay: (castor.cse.buffalo.edu) X-PM-EL-Spam-Prob: : 7% Message-ID: <200704171257.l3HCvc7l022348@castor.cse.Buffalo.EDU> Date: Tue, 17 Apr 2007 08:57:38 -0400 Reply-To: "William J. Rapaport" Sender: "CVA Seminar, Spring 2007" From: "William J. Rapaport" Subject: Re: CSE727: Comments on Hastings & Lytinen 1994 To: CSE727-SP07-LIST@LISTSERV.BUFFALO.EDU Precedence: list List-Help: , List-Unsubscribe: List-Subscribe: List-Owner: List-Archive: X-UB-Relay: (castor.cse.buffalo.edu) X-DCC-Buffalo.EDU-Metrics: castor.cse.Buffalo.EDU 1029; Body=0 Fuz1=0 Fuz2=0 X-Spam-Status: No, score=-2.5 required=5.0 tests=AWL,BAYES_00 autolearn=unavailable version=3.1.7 X-Spam-Checker-Version: SpamAssassin 3.1.7 (2006-10-05) on ares.cse.buffalo.edu X-Virus-Scanned: ClamAV 0.88.6/3110/Tue Apr 17 07:57:27 2007 on ares.cse.buffalo.edu X-Virus-Status: Clean Status: R Content-Length: 1147 Joe wrote: | ...I wondered what the effects of combining the CVA algorithms with | such detailed sets of domain knowledge (such as the military domain given in | the paper) would be. ... | | On the CVA side of things, one of the proposed seminar projects was to | integrate CYC with SNePS to provide a very large set of background | knowledge. This is, I think, related to Yana's question. One of the things I hope you all have come to realize while doing your project is how much prior knowledge is needed to do CVA! If Schatz & Baldwin and Beck et al. have any point at all, it's that the textual context alone is probably *not* enough to do good CVA. The more PK, the better. And this, in turn, suggests that CVA is a good candidate for being an "AI-complete" problem, i.e., an AI problem that requires the solution of all other AI problems for its own solution. For a little bit more on this notion, see Shapiro, Stuart C. (1992), "Artificial Intelligence", in Stuart C. Shapiro (ed.), Encyclopedia of Artificial Intelligence, second edition (New York: John Wiley & Sons): 54-57. http://www.cse.buffalo.edu/~shapiro/Papers/ai.ps From owner-cse727-sp07-list@LISTSERV.BUFFALO.EDU Tue Apr 24 19:25:46 2007 Received: from ares.cse.buffalo.edu (ares.cse.Buffalo.EDU [128.205.32.79]) by castor.cse.Buffalo.EDU (8.13.6/8.12.10) with ESMTP id l3ONPjOU000188 for ; Tue, 24 Apr 2007 19:25:45 -0400 (EDT) Received: from front2.acsu.buffalo.edu (upfront.acsu.buffalo.edu [128.205.4.140]) by ares.cse.buffalo.edu (8.13.6/8.13.6) with SMTP id l3ONPhmd039323 for ; Tue, 24 Apr 2007 19:25:43 -0400 (EDT) Received: (qmail 25515 invoked from network); 24 Apr 2007 23:25:43 -0000 Received: from mailscan8.acsu.buffalo.edu (128.205.7.55) by front2.acsu.buffalo.edu with SMTP; 24 Apr 2007 23:25:43 -0000 Received: (qmail 25324 invoked from network); 24 Apr 2007 23:25:40 -0000 Received: from deliverance.acsu.buffalo.edu (128.205.7.57) by front2.acsu.buffalo.edu with SMTP; 24 Apr 2007 23:25:40 -0000 Received: (qmail 11432 invoked from network); 24 Apr 2007 23:25:37 -0000 Received: from listserv.buffalo.edu (128.205.7.35) by deliverance.acsu.buffalo.edu with SMTP; 24 Apr 2007 23:25:37 -0000 Received: by LISTSERV.BUFFALO.EDU (LISTSERV-TCP/IP release 14.5) with spool id 5008963 for CSE727-SP07-LIST@LISTSERV.BUFFALO.EDU; Tue, 24 Apr 2007 19:25:37 -0400 Delivered-To: CSE727-SP07-LIST@LISTSERV.BUFFALO.EDU Received: (qmail 2825 invoked from network); 24 Apr 2007 23:25:36 -0000 Received: from mailscan8.acsu.buffalo.edu (128.205.7.55) by listserv.buffalo.edu with SMTP; 24 Apr 2007 23:25:36 -0000 Received: (qmail 9747 invoked by uid 60001); 24 Apr 2007 23:25:35 -0000 X-Mailer: University at Buffalo WebMail Cyrusoft SilkyMail v1.1.11 MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 8bit X-Originating-IP: 128.205.208.233 X-UB-Relay: (internal) X-PM-EL-Spam-Prob: : 7% Message-ID: <1177457135.462e91ef50e8a@mail3.buffalo.edu> Date: Tue, 24 Apr 2007 19:25:35 -0400 Reply-To: ask8@BUFFALO.EDU Sender: "CVA Seminar, Spring 2007" From: Ashish Kulkarni Subject: CSE 727: Hastings and Lytinen To: CSE727-SP07-LIST@LISTSERV.BUFFALO.EDU Precedence: list List-Help: , List-Unsubscribe: List-Subscribe: List-Owner: List-Archive: X-UB-Relay: (deliverance.acsu.buffalo.edu) X-DCC-Buffalo.EDU-Metrics: castor.cse.Buffalo.EDU 1335; Body=0 Fuz1=0 Fuz2=0 X-Spam-Status: No, score=-2.3 required=5.0 tests=AWL,BAYES_00 autolearn=unavailable version=3.1.7 X-Spam-Checker-Version: SpamAssassin 3.1.7 (2006-10-05) on ares.cse.buffalo.edu X-Virus-Scanned: ClamAV 0.88.6/3157/Tue Apr 24 16:25:13 2007 on ares.cse.buffalo.edu X-Virus-Status: Clean Status: R Content-Length: 1279 By looking at the representation Camille uses to learn new words, I find that there was a good connections made when it came to connecting nodes to representing a few sentences, but I am wondering how the system would work if this network is placed among a much bigger network of sentences where a lot of lines are represented and lots of words are used differently. For example two representations are made in which one of them talked about terrorists attacking food, and the second one representing hungry children attacking the food when their mother cooks it. Now, if the two scenarios are combined, there will be just one node for attack, and it would contradict the meanings with different approach. If this would confuse Camille, then should the belief revision system (if there is any) create two nodes ‘attack’ meaning different. Would Camille or any software in that case be smarter if it has a humongous background knowledge of its own, like humans have. Because as mentioned on page 400 under ‘Psychological connections’ in article part b, children learn nouns well before and faster than they learn verbs. This shows that humans learn the meaning of the words first before they learn to link it. Would trying something similar work better of programs like Camille? From owner-cse727-sp07-list@LISTSERV.BUFFALO.EDU Sun Apr 29 21:36:34 2007 Received: from ares.cse.buffalo.edu (ares.cse.Buffalo.EDU [128.205.32.79]) by castor.cse.Buffalo.EDU (8.13.6/8.12.10) with ESMTP id l3U1aYs0016590 for ; Sun, 29 Apr 2007 21:36:34 -0400 (EDT) Received: from front1.acsu.buffalo.edu (coldfront.acsu.buffalo.edu [128.205.6.89]) by ares.cse.buffalo.edu (8.13.6/8.13.6) with SMTP id l3U1aU8i039659 for ; Sun, 29 Apr 2007 21:36:30 -0400 (EDT) Received: (qmail 7668 invoked from network); 30 Apr 2007 01:36:30 -0000 Received: from mailscan1.acsu.buffalo.edu (128.205.6.133) by front1.acsu.buffalo.edu with SMTP; 30 Apr 2007 01:36:30 -0000 Received: (qmail 22821 invoked from network); 30 Apr 2007 01:36:30 -0000 Received: from deliverance.acsu.buffalo.edu (128.205.7.57) by front3.acsu.buffalo.edu with SMTP; 30 Apr 2007 01:36:30 -0000 Received: (qmail 5598 invoked from network); 30 Apr 2007 01:36:19 -0000 Received: from listserv.buffalo.edu (128.205.7.35) by deliverance.acsu.buffalo.edu with SMTP; 30 Apr 2007 01:36:19 -0000 Received: by LISTSERV.BUFFALO.EDU (LISTSERV-TCP/IP release 14.5) with spool id 5143271 for CSE727-SP07-LIST@LISTSERV.BUFFALO.EDU; Sun, 29 Apr 2007 21:36:19 -0400 Delivered-To: cse727-sp07-list@listserv.buffalo.edu Received: (qmail 8230 invoked from network); 30 Apr 2007 01:36:19 -0000 Received: from mailscan3.acsu.buffalo.edu (128.205.6.135) by listserv.buffalo.edu with SMTP; 30 Apr 2007 01:36:19 -0000 Received: (qmail 24603 invoked from network); 30 Apr 2007 01:36:18 -0000 Received: from mta9.adelphia.net (68.168.78.199) by smtp4.acsu.buffalo.edu with SMTP; 30 Apr 2007 01:36:18 -0000 Received: from [127.0.0.1] (really [98.1.118.198]) by mta9.adelphia.net (InterMail vM.6.01.05.02 201-2131-123-102-20050715) with ESMTP id <20070430013617.DJRW14403.mta9.adelphia.net@[127.0.0.1]> for ; Sun, 29 Apr 2007 21:36:17 -0400 User-Agent: Thunderbird 1.5.0.10 (Windows/20070221) MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-UB-Relay: (mta9.adelphia.net) X-PM-Spam-Prob: : 7% Message-ID: <46354843.3000408@cse.buffalo.edu> Date: Sun, 29 Apr 2007 21:37:07 -0400 Reply-To: Jeffrey Howell Sender: "CVA Seminar, Spring 2007" From: Jeffrey Howell Subject: CSE 727: Hastings & Lytinen (1994ab) To: CSE727-SP07-LIST@LISTSERV.BUFFALO.EDU Precedence: list List-Help: , List-Unsubscribe: List-Subscribe: List-Owner: List-Archive: X-UB-Relay: (mta9.adelphia.net) X-DCC-Buffalo.EDU-Metrics: castor.cse.Buffalo.EDU 1335; Body=0 Fuz1=0 Fuz2=0 X-Spam-Status: No, score=-2.6 required=5.0 tests=AWL,BAYES_00 autolearn=unavailable version=3.1.7 X-Spam-Checker-Version: SpamAssassin 3.1.7 (2006-10-05) on ares.cse.buffalo.edu X-Virus-Scanned: ClamAV 0.88.6/3179/Sun Apr 29 06:28:45 2007 on ares.cse.buffalo.edu X-Virus-Status: Clean Status: R Content-Length: 1396 The comparison of human acquisition of noun meanings versus verb meanings is fascinating. As so often seems to happen, though, I find the most interesting bit of information is an aside.. in this case, the footnote in "Objects, Actions, Nouns, and Verbs" on page 400 that mentions the difference in ease of acquisition of the different parts of speech in different languages. The implication that the relative speed at which different vocabulary is learned is at least partially dependent on the structure of the language rather than being 'hard-coded' in the brain is an important one. If nothing else, it seems that syntax plays a role in vocabulary acquisition as well as semantics. Back to the thrust of the articles - a lot of the weaknesses I perceived in Camille are commented on by the authors in both articles, and probably aren't worth repeating here. A definite problem I see that *wasn't* commented on very much is the fact that this system is both parsing the syntax of the sentences and examining the semantics. As a final goal, this is obviously the ideal, but when neither part of the whole is fully understood, I wonder if attempting to make both work all at the same time might not be a bit much to bite off - Camille has enough room for improvement just based on her semantic interpretations without throwing in the problem of parsing natural language!