Argumentation in Computer Media - Pittsburgh Interest Group Newsletter - 25 March 1999 - updated 2 April 99 Contents: 1. Summary of 24 March 1999 meeting 2. Next meeting - Thursday, April 8, 4:45-6 pm - 4625 Wean Hall, CMU 3. Meeting date/time preference survey 4. Distribution of newsletter, announcements 5. To be added to the mailing list 6. Announcement about CHI99 SIG ======================================================================= 1. Summary of 24 March 1999 meeting There were 7 participants, affiliated (currently or in the recent past) with Pitt/LRDC (Education, ISP) and CMU (SCS, English, and Psychology). Two people sent word that they would like to have come but couldn't make it. Informal presentations continued on the participants' perspectives on argumentation based on their own research, including * academic discourse -- teaching CMU students how to analyze and construct written arguments (using the Toulmin model, which was based on legal arguments) -- this prompted discussion on the difference between "persuasion" (e.g. a Nike shoe ad) and "argumentation", which follows rules that are accepted by a particular group (science, law) * computer-supported science education -- use of graphical representations of argument structure (using the Toulmin model) to represent the "logic" of scientific arguments in order to teach college students about what constitutes a valid scientific argument -- this prompted discussion on the possible effectiveness of such an approach, not just for teaching, but for use by authors in presenting arguments in computer media. I.e., why not use the computer to represent the structure of an argument when presenting the argument? Some of us brought up the cost to the audience of learning to interpret the argument representation scheme, and the potential problem of making a complex argument more difficult to comprehend because of the additional information. Also, some argued that a static synchronous representation of the argument would fail to capture the effect of presenting elements of an argument in a particular order. Someone brought up the example of how in complex papers in philosophy, presentation order is critical. On the other hand, the computer medium could be designed to represent temporal dependencies and hide information about argument structure unless needed. * computer-supported science education -- a simpler version of the above targeted at middle-school students (distinguishes hypotheses from data and the two relations of support-for and support-against) - the computerized representation supports collaboration among students and enables the students' arguments to be syntactically checked for correctness * computer-supported argumentation for newsgroups - a system (Hermes) used to classify items posted to an on-line newsgroup in terms of the role of the posting in the on-going debate * computer-generated evaluative arguments tailored to a model of the user's preferences, e.g., the system recommends a certain house for sale given a model of the user's preferences about price, location, etc. - uses decision-theory to compute the strengths of judgements in the user model, which is used in selecting the content and organization of the arguments and for selection of evaluative words (e.g. "expensive" versus "cheap") ============================================================================ 2. Next meeting Date: Thursday, April 8, 1999 Time: 445-5 pm socializing, 5-630 pm discussion Location: 4625 Wean Hall, Carnegie Mellon University Agenda: Discussion. Let's classify the various perspectives on argumentation that we have shared in the last two meetings. ========================================================================== 3. Date/time/preferences: [Thanks to those who have already responded to this!] I will try to take your preferences into consideration when scheduling future meetings. Please help me cope with the amount of email this may generate by standardizing your reply as follows. Please detach the following questionnaire and return it to me by email (ascii only) with your answers or you can give me this information in person at the next meeting. Thanks! ------- detach survey here ---------------------------------------- To indicate your preferences for meeting times, please please mark the cells of this matrix as: very good (++), good (+), ok (-), bad (--), or other (leave blank). The vertical axis indicates day of the week and month, e.g., "Tue Mar" means Tuesdays in March. The horizontal axis shows meeting time (p.m.). 4-530 430-6 5-630 530-7 Tue Mar Wed Mar Thu Mar Tue Apr Wed Apr Thu Apr Tue May Wed May Thu May Miscellaneous comments on date/time: At a future meeting, I would like to volunteer to give a longer presentation on: ---------- end of survey ---------------------------------------- ======================================================================= 4. Distribution of newsletter, announcements Feel free to forward this to colleagues who may be interested. I have discontinued sending news of the group to general-purpose local mailing lists because of complaints that I received. ======================================================================= 5. To be added to the mailing list: If someone forwarded this to you and you would like to be added to the mailing list please send your email address to me: nancy.green@cs.cmu.edu ======================================================================= 6. Announcement: CHI99 SIG on Natural Language in Computer-Human Interaction Although there are several annual conferences that provide an opportunity for the Natural Language Processing (NLP) and CHI communities to interact, such as the yearly International Conference on Intelligent User Interfaces (IUI) and the Annual ACM Symposium on User Interface Software and Technology (UIST), the time is right for interaction at their respective primary meetings, the annual conference of the Association for Computational Linguistics (ACL) and the annual CHI conference. The primary goal of this SIG is to provide an opportunity for CHI attendees from both the CHI and NLP research communities who are interested in use of NL in the computer interface to discuss issues of mutual interest. Some of the technical issues of interest to both communities are 1. For what and under what conditions is NL appropriate in the human-computer interface? For what types of tasks or communication is NL appropriate? How does modality (such as spoken, typewritten, and direct manipulation of pre-generated text) affect appropriateness? How does the appropriateness of NL in computer media differ from its appropriateness in traditional forms of communication (such as face-to-face conversation and print media)? How do performance limitations of automated NL interpretation and generation technologies (e.g., speech recognition errors) affect appropriateness? What approaches can be used to answer the above? 2. What are the unique technical requirements for NLP in the human-machine interface? (e.g., coordination of generated text and graphics, incremental generation and interpretation, robust interpretation, modeling turn-taking and initiative in dialogue) For more information see our web site: http://www.cs.cmu.edu/~sage/CHI99SIG.html