Natural Language, Knowledge Representation, 
and Discourse 
James F. Allen and Lenhart K. Schubert 
Department of Computer Science 
University of Rochester 
Rochester, NY 14627 
Goals 
The principal objective of this project is to develop a sys- 
tem for representing and reasoning about the discourse con- 
text in extended man-machine dialogs. Current focus areas 
include the development of a general theory of multi- agent 
planning to account for the structure of natural-langnage 
dialog, the development of a general knowledge represen- 
tation for capturing a wide range of natural language 
semantics, and the development of a general, error- tolerant 
parser and semantic interpreter for English that can be 
guided by discourse information. Specifically, we are 
developing a model of discourse plans that includes actions 
such as introducing a new topic, as well as the actions of 
clarifying, correcting or acknowledging parts of the pre- 
vious dialog. We are explonng how far the planning ap- 
proach can be extended, and how the "traditional" language 
components, i.e. parsing, semantic interpretation and dis- 
course processing, relate to the planning component. 
Recent Results 
We developed a model of indirect speech act interpretation 
that has the generality plan-based approaches, but is sen- 
sitive to the syntactic form of the sentence (Hinkelman & 
Allen, 1989). We also demonstrated an initial version 
ECOLOGIC, for representing and reasoning about nar- 
rative events, using commonsense uncertain knowledge. 
Most recently, we designed and implemented a detailed 
theory of tense and aspect that addresses well-known 
problems in the systematic determination of "reference 
times" for tense-aspect constructs. This is reported in these 
proceedings. 
Plans for the coming year 
The major initiative this year is the construction of a 
prototype NL dialog system operating in a simple, but 
realistic task domain requiring considerable man-machine 
interaction. This domain is one of scheduling transportation 
actions in a complex (simulated) world where only partial 
knowledge of the world state can ever be obtained. The 
system must maintain as accurate a representation of the 
world as possible and keep the human informed, as well as 
assisting the human in constructing, monitoring and debug- 
ging the transportation plans. Within the next year, we 
plan to have a dialog system that can understand and take 
part in an extended dialog involving clarification, elabora- 
tion and confirmation before the final plan is agreed upon 
and executed. Related to this project, we are starting a 
project on the use of prosodic cues for signaling focus in 
speech act interpretation. 
We also plan to extend ECOLOGIC to allow for "adjudica- 
tive inference" (arriving at an unambiguous interpretation 
using multiple, possibly conflicting sources of knowledge), 
and "narrative inference" (the inference of identity between 
discourse entities, and of causal connections based on nar- 
rative order). Finally, we plan to develop a better theoreti- 
cal foundation for wh-question answering, which will use 
all of the existing rules of interpretation (including syntac- 
tic parsing, logical form generation, and de-indexicalization 
rules) "in reverse". 
429 
