[comp.ai] IJCAI workshop on natural language generation

hovy@isi.edu (Eduard Hovy) (03/27/91)

                      ************************
                      *    Call For Papers   *
                      ************************


                         IJCAI 1991 Workshop
                        August 24 - 25, 1991


           Decision Making throughout the Generation Process


While there is general agreement that Natural Language Generation (NLG)
is a largely deterministic decision making process, there is little
agreement between researchers on the parameters of this process. 
Different researchers have focused on different parts of the generation
process, such as content planning, text organization and linguistic
realization, and taken into account different kinds of knowledge, such
as discourse structure, user models and taxonomic domain knowledge.
Hence, their decision spaces, while plainly overlapping in function, are
incomparable in their details.  This workshop will attempt to address
these differences concretely by bringing together researchers in NLG and
related fields, and encouraging them to present their work specifically
in terms of the decision making processes undertaken by their systems. 

In order to be able to describe the generation process, we need to
arrive at  some consensus as to the space of decisions in the process
overall. This, in turn, will allow us to better understand and take
advantage of each other's work, since researchers will be able to
identify clearly which portion of the decision space their work is
intended to cover. In this context, an important question that must be
answered is whether the different decision types have any commonalities
that can be exploited; for example, could the reasoning processes that
determine text planning choices also be useful for lexical selections?
In addition, as indicated above, at present there is little agreement
with respect to the answers to the following questions: What are the
decisions made by a text generation system? How should alternatives be
represented? What control structure determines the order in which the
decisions should be made? and What is the effect of a decision on
subsequent decisions? 

The identification of the decisions encountered during the text
generation process is an essential step towards the eventual formal
specification of this process and the development of evaluation metrics
for it. In addition, it allows practitioners in the field to better
understand and take advantage of each other's work, as it enables them
to determine which portion of the decision space a particular system or
procedure is intended to cover.  



Submissions

We encourage papers in areas of NLG including (but not limited to):
content selection, text organization, utterance composition, reactive
explanation, cooperative responses, lexical choice, stylistics and
linguistic realization. Papers should address the above questions, argue
for a particular approach, or describe a particular system from the
point of view of decision making.   

Interested participants should submit 6 copies of a 5 page abstract and
a brief description of their research activities by May 1, 1991. The
following information should be included: name, mailing address, phone
number and electronic mail address.  

Submissions may be made (in order of preference) via Real-mail, FAX, or
electronic mail (LaTeX article style or directly printable form) to
either co-chair:

Dr. Marie Meteer (co-chair), BBN, 10 Moulton St., Cambridge, MA 02138,
mmeteer@bbn.com, FAX:  (617) 873-3776

Dr. Ingrid Zukerman (co-chair), Dept. of Computer Science, Monash 
University, Clayton, VICTORIA 3168, AUSTRALIA,
ingrid@bruce.cs.monash.oz.au, FAX: (your international operator code)
(613) 565-5146



Schedule

 May 1, 1991 	Submissions due 
 June 1, 1991 	Notification of acceptance   
 July 1, 1991   Camera-ready copies due
 July 15	Workshop registration fee ($US65) due to IJCAI
 Aug 24-25 	Workshop



Organizing and Program Committee

 Dr. Marie Meteer (co-chair), BBN 
 Dr. Ingrid Zukerman (co-chair), Computer Science Dept., Monash University  
 Dr. Eduard Hovy, USC/ISI    
 Dr. David McDonald, Content Technologies   
 Dr. Cecile Paris, USC/ISI