nl-kr-request@CS.RPI.EDU (NL-KR Moderator Chris Welty) (04/17/91)
NL-KR Digest (Tue Apr 16 13:35:33 1991) Volume 8 No. 18 Today's Topics: speech seminar series NLP Databases in Chinese commercial NLIs IJCAI91 CILS Calendar Submissions: nl-kr@cs.rpi.edu Requests, policy: nl-kr-request@cs.rpi.edu Back issues are available from host archive.cs.rpi.edu [128.213.10.18] in the files nl-kr/Vxx/Nyy (ie nl-kr/V01/N01 for V1#1), mail requests will not be promptly satisfied. If you can't reach `cs.rpi.edu' you may want to use `turing.cs.rpi.edu' instead. BITNET subscribers: we now have a LISTSERVer for nl-kr. You may send submissions to NL-KR@RPIECS and any listserv-style administrative requests to LISTSERV@RPIECS. ----------------------------------------------------------------- To: nl-kr@cs.rpi.edu Date: Mon, 15 Apr 91 15:16:51 EDT >From: Helene George <hgeorge@bbn.com> Subject: speech seminar series Speech Seminar Series Who: Dr. Richard A. Bolt, MIT Media Laboratory What: SPEECH, GESTURE, AND GAZE IN MULTI-MODAL NATURAL DIALOG When: Wednesday, April 24, 1:30pm Where: 4th floor large conference room, 10 Moulton Street (6/471) Abstract: People converse with each other primarily through combinations of speech and gesture, while paying attention to the other's gaze. Might someday a person deal similiarly with a computer and be as readily understood? Such dialog requires technologies to capture human speech, gesture, and looking, all of which means exist in useful, if not ideal, forms today. Equally necessary is the program intelligence to interpret such outputs, infer the human's intent, and generate an appropriate response in graphics and/or sound. Richard A. Bolt is a Senior Research Scientist at the MIT Media Lab, and Director of the Lab's Advanced Human Interface Group. He has lectured nationally and internationally on human/computer interaction, and is author of THE HUMAN INTERFACE published by Van Nostrand Reinhold. ------------------------------ To: nl-kr@cs.rpi.edu >From: LUAKT%nusdiscs.bitnet@cs.rpi.edu Date: Sat, 13 Apr 91 16:12 H Subject: NLP Databases in Chinese To: nl-kr@cs.rpi.edu Subject: Re: NL-KR Digest, Volume 8 No. 15 RE: NLP Databases Personally I use two databases of Chinese characters. One is a database of 46,000 Chinese words, with Hanyu Pinyin and frequencies of occurrence. This is availabe at US$250 from Chinese and Oriental Languages Information Society. The second is a personal database for Chinese word semantic classes. It classify 70,000 Chinese words into 12 major, 94 medium and 1428 minor classes. Lua K T > lua kim teng (LUAKT@NUSDISCS.BITNET) > National University of Singapore > Department of Information Systems and Computer Science ------------------------------ To: nl-kr@cs.rpi.edu >From: imlah@canon-research-europe.co.uk (Bill Imlah) Date: Fri, 12 Apr 91 15:55:28 GMT Subject: commercial NLIs In comp.ai.nlang-know-rep you write: >Can anyone supply me with names and addresses of companies >supplying natural language data base interfaces commercially? NLI aka DATATALKER: Natural Language Inc 1802 5th St Berkeley Cal. 94710 (415) 841 3500 I know of a system in development called BIMloqui, I think, from BIM. Try dani%sunbim.be@ukc.ac.uk for more info. I have also heard of "Spock" (US company) and Batelle but know nothing about them. Spock comes from: Dynamics Research Corporation Dept 948 60 Frontage Road Andover MA 01810 tel. 508/475-9090 I would be interested in knowing what you find out about available NLIs. Will you be posting? > buy something "off the shelf" and plug in a dictionary > to use it with a data base they have. Don't expect too much from current commercialised technology. "plugging in" is likely to be very effort-intensive (and probably frustrating) - ------------------------------------------------------- Bill Imlah imlah@canon.co.uk Canon Research Centre Europe, 17/20 Frederick Sanger Rd. The Surrey Research Park, Guildford, Surrey, GU2 5YD, UK. ------------------------------ To: nl-kr@cs.rpi.edu Date: Thu, 11 Apr 91 16:28:53 EDT >From: morrow@das.harvard.edu Subject: IJCAI91 ========================================================================== 12th INTERNATIONAL JOINT CONFERENCE ON ARTIFICIAL INTELLIGENCE 24 - 30 August, 1991 * Darling Harbour * Sydney * Australia * IJCAI-91 ======== IJCAI-91 will be held at Darling Harbour, Sydney, Australia, 24-30 August, 1991. Professor Barbara Grosz of Harvard University is the Conference Chair; Professor John Mylopoulos and Professor Ray Reiter, both of the University of Toronto, are Program CoChairs; and Professor Michael McRobbie of the Australian National University is the Australian National Committee Chair. Dr. Donald Walker of Bellcore is Secretary-Treasurer for the Conference. REGISTRATION BROCHURES AND FORMS FOR IJCAI-91 ============================================= The Australian National Committee has arranged for Registration Brochures and Forms for IJCAI-91, to be obtained by three methods: hard copy, ftp, e-mail. These are described below. As noted below, the e-mail option is an option of last resort only. ******************************************************************************* IJCAI-91 WORKSHOP PROGRAM ========================= It is strongly recommended that people interested in the IJCAI-91 Workshop Program obtain a copy of the Registration Brochure via anonymous ftp (see below) as soon as possible since the closing date for submissions for all workshops is 15 May, or in some cases earlier. ******************************************************************************* OBTAINING REGISTRATION BROCHURES ================================ 1. Hard Copies - ------------- Hard copies of the Registration Brochures will be available mid- to late April, 1991. Brochures will then be mailed to all current members of the national and regional AI societies and interest groups in the following countries: Australia, Canada, China, Europe (to all who receive AICOM publications), Japan, Korea, Mexico, United States (AAAI), USSR, and to all people who have requested information from IJCAI-91 Conference Committee members or the IJCAI-91 Secretariat. Attempts are being made to send mail to Eastern European countries as well. Anyone who is not a member of one of these societies or does not receive a brochure by mid-May can request one by sending an e-mail message to: ijcai.rego.hardcopy@vulcan.anu.edu.au PLEASE NOTE THAT THIS IS AN E-MAIL ADDRESS TO WHICH *ONLY* E-MAIL REQUESTS FOR *HARD COPIES* OF THE REGISTRATION BROCHURES AND FORMS SHOULD BE SENT. IT IS *NOT* A GENERAL IJCAI CORRESPONDENCE ADDRESS. Requests for hard copies of the Registration Brochure and Form and any other enquiries concerning IJCAI-91, can be made directly to: IJCAI-91 Secretariat PO Box 787 Potts Point, NSW 2011 AUSTRALIA Tel: (+61-2) 3572600 Fax: (+61-2) 3572950 2. Anonymous ftp - --------------- For those connected to the Internet, the IJCAI Registration Brochure and Form are available as compressed POSTSCRIPT files via anonymous ftp from vulcan.anu.edu.au (IP number 130.56.4.173) (in /IJCAI) and from uunet.uu.net (alias ftp.uu.net, IP number 137.39.1.2) (in /doc/IJCAI) ********************************************************************** NOTE: So as to reduce unnecessary network traffic and the number of transfers from vulcan.anu.edu.au, please make the relevant POSTSCRIPT files publicly available at your site. Please contact your local system manager for details of how best to do so. ********************************************************************** To transfer the Registration Brochure and Form proceed as follows: diana 201 $ ftp vulcan.anu.edu.au Connected to vulcan.anu.edu.au. 220 vulcan FTP server ($Header: ftpd.c 2.5 89/12/15 $) ready. Name (vulcan.anu.edu.au:gustav): anonymous 331 Guest login ok, send ident as password. Password: Now type in your full e-mail address as password (it will be logged). After a successful login change to directory IJCAI. You can issue the command "ls" to see the contents of that directory: 230 Guest login ok, access restrictions apply. ftp> cd IJCAI (on uunet.uu.net do "cd pub/IJCAI" instead) ftp> ls 200 PORT command successful. 150 Opening data connection for /bin/ls (ascii mode) (0 bytes). README reg.brochure.A4.ps.Z reg.brochure.11.ps.Z reg.form.A4.ps.Z reg.form.11.ps.Z 226 Transfer complete. 64 bytes received in 0.039 seconds (1.6 Kbytes/s) ftp> Now enter ftp commands "binary" (in order to transfer the compressed files which are not ASCII) and "prompt" (in order to be able to transfer more than one file with one command without being molested by ftp with unnecessary questions): ftp> binary 200 Type set to I. ftp> prompt Interactive mode off. ftp> At this stage things depend on whether your printers use 11 inch paper (e.g. as in the USA) or A4 size paper. If you use 11 inch paper say: ftp> mget README *.11.* After a successful transfer, quit ftp and unpack the files: ftp> quit 221 Goodbye. % uncompress reg.brochure.11.ps.Z reg.form.11.ps.Z The two files: "reg.brochure.11.ps" and "reg.form.11.ps" can be now sent directly to any PostScript printer, e.g., % lpr -Plsr -s reg.brochure.11.ps % lpr -Plsr -s reg.form.11.ps After uncompressing file reg.brochure.11.ps will be about 3.5 Mbytes in size. For this reason you should use "-s" option while invoking lpr as lpr will then make a symbolic link between that file and the spooler rather than copy the file to the spooler. If your printer uses A4 instead of 11 inch paper repeat the above procedure for the files "reg.brochure.A4.ps.Z" and "reg.form.A4.ps.Z". On some Sys V systems, you may not be able to create files with names longer than 10 or so characters. In that case you may have to transfer files one by one, e.g.: ftp> get reg.brochure.11.ps.Z bro.ps.Z ... ftp> get reg.form.11.ps.Z form.ps.Z ... Also, on Sys V systems the name of the print command is "lp" and some versions of Sys V do not support symbolic links. However, there should still be an option to lp which will prevent the printed file from being copied to the spooler. In case of problems or difficulties contact your local system manager or: ijcai.network.support@vulcan.anu.edu.au. 3. By automatic e-mail - --------------------- ****************************************************************************** NOTE: This procedure is complex and it puts a great burden on any system that participates in the mailing process. Use it ONLY AS A LAST RESORT. If you transfer files in this way, make sure that they become publicly accessible within your site and its neighbourhood. ****************************************************************************** For those not connected to the Internet, A4 and 11'' versions of the IJCAI Registration Brochure and Form can be obtained via e-mail in uuencoded compressed shar format. Send a message to ijcai.rego.email.11@vulcan.anu.edu.au for the 11 inch documents and to ijcai.rego.email.A4@vulcan.anu.edu.au for the A4 size documents. The corresponding aliases on vulcan will automatically send you back several shar files. For 11 inch documents you will receive: 94 r.b.11.ps.Z.shar.1 94 r.b.11.ps.Z.shar.7 66 r.b.11.ps.Z.shar.10 94 r.b.11.ps.Z.shar.8 94 r.b.11.ps.Z.shar.2 94 r.b.11.ps.Z.shar.9 94 r.b.11.ps.Z.shar.3 94 r.b.11.ps.Z.shar.1 94 r.b.11.ps.Z.shar.4 94 r.f.11.ps.Z.shar.2 94 r.b.11.ps.Z.shar.5 18 r.f.11.ps.Z.shar.3 94 r.b.11.ps.Z.shar.6 and for the A4 size documents: 94 r.b.A4.ps.Z.shar.1 94 r.b.A4.ps.Z.shar.6 94 r.b.A4.ps.Z.shar.10 94 r.b.A4.ps.Z.shar.7 9 r.b.A4.ps.Z.shar.11 94 r.b.A4.ps.Z.shar.8 94 r.b.A4.ps.Z.shar.2 94 r.b.A4.ps.Z.shar.9 94 r.b.A4.ps.Z.shar.3 94 r.f.A4.ps.Z.shar.1 94 r.b.A4.ps.Z.shar.4 94 r.f.A4.ps.Z.shar.2 94 r.b.A4.ps.Z.shar.5 18 r.f.A4.ps.Z.shar.3 Observe the numbers in front of file names. These are the sizes of the files in kilobytes. ****************************************************************************** WARNING: Do not send an automatic e-mail request if: 1) you do not have enough space in the mail spooler to accomodate these files; 2) your mail links are very slow; 3) your mailer cannot process files of this size; 4) you can use ftp from a machine to which you have an access. ****************************************************************************** To unpack the documents, save the messages on the files with the corresponding names in a separate directory. For example, call that directory IJCAI. If you have program unshar (the latest versions of shar and unshar can be obtained from volume 15 of comp.sources.unix) then do: $ cd IJCAI $ unshar * otherwise you will have to strip the mail headers from all files by hand and then run /bin/sh on them (Bourne shell syntax assumed): $ for i in * > do > /bin/sh $i > done This will create numerous files in your directory with extensions .uu.aa, .uu.ab, .uu.ac, etc. It will also create two command files: r.b.A4.ps.Z.unpack r.f.A4.ps.Z.unpack (if you have asked for A4 size documents). Execute these to finally recreate the PostScript files r.b.A4.ps (the registration brochure) and r.f.A4.ps (the registration script) The two files: "r.b.A4.ps" and "r.f.A4.ps" can be now sent directly to any PostScript printer, e.g., % lpr -Plsr -s r.b.A4.ps % lpr -Plsr -s r.f.A4.ps The file r.b.A4.ps is about 3.5 Mbytes in size. For this reason you should use the "-s" option while invoking lpr as lpr will then make a symbolic link between that file and the spooler rather than copy the file to the spooler. If your printer uses 11 inch instead of A4 paper repeat the above procedure for the files "r.b.11.ps.Z" and "r.f.11.ps.Z". In case of problems or difficulties contact your local system manager or: ijcai.network.support@vulcan.anu.edu.au. PLEASE NOTE THAT THESE ARE E-MAIL ADDRESSES TO WHICH *ONLY* E-MAIL REQUESTS FOR THE REGISTRATION BROCHURE AND REGISTRATION FORM *IN UUENCODED COMPRESSED SHAR FORMAT* SHOULD BE SENT. THESE ARE *NOT* GENERAL IJCAI CORRESPONDENCE ADDRESSES. ******************************************************************************* * A note to system managers: * * * * The shar files described above live in ftp anonymous area on * * vulcan.anu.edu.au (IJCAI/shar). You are welcome to transfer these files * * to your system via ftp and set up a similar facility to the one described * * above. A script which bounces mail lives on "bouncepost" in the same * * directory. A script "pack" in that directory illustrates the packing * * process, in case you have to unpack files by hand. * ******************************************************************************* ------------------------------ To: nl-kr@cs.rpi.edu Subject: CILS Calendar X-Mailer: MH 6.6 #5[UCI] Date: Mon, 15 Apr 91 16:48:08 -0500 >From: colleen@tira.uchicago.edu _________________ T H E C I L S C A L E N D A R ________________ The Center for Information and Language Studies Joseph Regenstein Library, University of Chicago, Chicago, IL 60637 Subscription requests to: cils@tira.uchicago.edu ____________________________________________________________________ Vol. 1, No. 23 April 15, 1991 ~*~ Upcoming events: 4/22 14:30 Ry 277 Lecture Lisa Rau, GE Research and Development 4/22 16:00 Wb 130 Workshop Greg Ward, Northwestern 4/29 14:30 Ry 277 Lecture Glenn Reid, RightBrain Software 5/6 14:30 Ry 277 Lecture David Lewis, CILS - ------------------------------ MONDAY, APRIL 22 Ry 277 Guest Lecture 2:30 p.m. Lisa Rau (rau@sol.crd.ge.com) (518) 387-5059; FAX: (518) 387-6845 Artificial Intelligence Laboratory GE Research and Development Center "CONCEPTUAL INFORMATION EXTRACTION AND RETRIEVAL" Abstract in last week's calendar. ***** 4:00 Workshop Wb 130 The Pragmatics of Language Greg Ward, Northwestern University For more information, please contact Jerrold Sadock, Department of Linguistics (2-8524, sadock@sapir) or Josef Stern, Department of Philosophy (2-8594, j06s@midway). The next speaker will be Stephen Schiffer, CUNY Graduate Center, on May 6. __________ MONDAY, APRIL 29 2:30 Glenn Reid, RightBrain Software Ry 277 (glenn%heaven.uucp@next.com) "The PostScript Distillery" Abstract in last week's calendar. __________ MONDAY, MAY 6 2:30 David D. Lewis (lewis@tira.uchicago.edu) Ry 277 Center for Information and Language Studies Text Classification: Statistical and Linguistic Issues Computer text classification systems include systems for the retrieval of documents in response to user queries, and systems for categorizing documents with respect to a set of meaningful categories. The recent resurgence of interest in information retrieval has been accompanied by a new emphasis on the use of natural language processing in text classification. In this talk we discuss two experiments on using syntactic analysis of natural language text to improve text retrieval performance. The importance of considering both linguistic and statistical concerns will be stressed in both cases. In the first study, syntactic parsing is used to extract indexing phrases from queries and documents. These phrases have better semantic properties than individual words, but inferior statistical properties. Feature clustering is then used to produce features which are superior on both semantic and statistical grounds, and which result in improved performance in a statistical retrieval system. The difficulty of comparing feature sets using conventional information retrieval test collections will be briefly discussed, as will an experiment which is in progress to evaluate this feature formation method on a text categorization task. In the second study, which is joint work with Bruce Croft (Univ. of Mass. at Amherst) and Howard Turtle (West Publishing), syntactic parsing of natural language queries is used in deciding how to structure a Bayesian inference net for document retrieval. This automatic analysis of natural language queries produced retrieval performance comparable to that obtained using hand structured queries, and is a promising approach to effective use of dependence models in text retrieval. - ----------- End of CILS Calendar ------------------------------ End of NL-KR Digest *******************