gency dispatcher, cooperating with the system to 
dynamically allocate resources to and make 
plans for solving problems as they arise in the 
world. The setting, Monroe County, NY, is con- 
siderably more complex than our previous do- 
mains (e.g. Pacifica, TRAINS), and raises new 
issues in knowledge representation and refer- 
ence. Emergencies include requests for medical 
assistance, car accidents, civil disorder, and 
larger problems such as flooding and snow 
storms. Resources at the user's disposal may 
include road crews, electric crews, ambulances, 
police units and helicopters. Some of the in- 
crease in mixed-initiative interaction comes 
from givi-n~ the_ system more knowledge of the 
tasks being solved. Some comes from the fact 
that the solution to one problem may conflict 
with the solution to another, either because of 
scheduling conflicts, scarce resources, or aspects 
of the physical world (e.g. an ambulance can't go 
down a road that has not been plowed). The 
range of tasks and complexity of the world allow 
for problem solving at different levels of granu- 
larity, making it possible for the system to take 
as much control over the task as the user per- 
mits. 
4. Important Contributions 
While a number of robust dialogue systems have 
been built in recent years, they mostly have op- 
erated in domains that require little if any rea- 
soning. Rather, the task is hard-coded into the 
system operation. One of the major goals of the 
TRIPS project has been to develop dialogue 
models and system architectures that support 
conversational interaction in domains where 
complex reasoning systems are required. One 
goal has been to build a fairly generic model in 
which different domains can then be specified 
fairly easily. On this front, we are seeing some 
success as we have now constructed versions of 
TRIPS in three different domains, and TRIPS° 
911 will be the fourth. In developing the system 
for new domains, the bulk of the work by far has 
been in system enhancements rather than in 
developing the domain models. 
The TRIPS-911 domain has forced a rethinking 
of the relationship between dialogue- 
management, problem-solving, the system's 
Figure 1: Monroe County map used in TRIPS-911 
own goal-pursuit and generation. The new ar- 
chitecture is designed to support research into 
mixed-initiative interactions, incremental gen- 
eration of content (in which the user might in- 
tervene before the system completes all it has to 
say), rich reference resolution models, and the 
introduction of plan monitoring and plan repair 
into the suite of plan management operations 
supported. The domain also can support longer 
and richer dialogues than in previous domains. 
More complex domains mean even more com- 
plex dialogues. The complexity arises from 
many factors. First, more complex dialogues 
will involve topic progression, development and 
resumption, and more complex referential phe- 
nomena. On the problem solving front, there will 
be more complex corrections, elaborations and 
modifications--forcing us to develop richer 
discourse models. In addition, the complexity of 
the domain demonstrates a need for better 
grounding behavior and a need for incremental 
dialogue-based generation. 
We have by no means solved these problems. 
Rather we have built a rich testbed, designed and 
implemented a plausible architecture, and have 
constructed an initial system to demonstrate 
basic capabilities in each of the problem areas. 
34 
5. Limitations 
TRIPS-911 is a first attempt at handling a do- 
main of this complexity. As such there are many 
capabilities that people have in such situations 
that are beyond the system's current capabilities. 
Some of the most important are: 
• Scale - we can only handle small domains 
and the existing techniques would not ex- 
tend directly to a realistic size 911 operation. 
To scale up we must face some difficult 
problems including reasoning about quanti- 
ties and aggregates, planning in large-scale 
domains (i.e., the real domains are beyond 
the capabilities of current plan technology), 
and performing intention recognition as the 
number of options increases. In addition, for 
an effective dialogue system, all this must be 
done in real-time. 
• Meta-talk - when faced with complex prob- 
lems, people often first generally discuss the 
problem and possible strategies for solving 
it, and later may explicitly direct attention to 
specific subproblems. The current TRIPS 
system does not support such discussion. 
• Time - in the 911 domain there are at least 
two temporal contexts that can be "used" by 
the conversants: there is the actual time (i.e., 
when they are talking), but there also is the 
time relative to a point of focus in a plan, or 
even simply talking about the past or the 
future. TRIPS-911 can currently interpret 
expressions with respect to the actual time. 
• Interleaved generation - when people are 
discussing complex issues, they often have 
to plan to communicate their content across 
several different utterances. There is no 
guarantee that the other conversant will not 
"interrupt" (e.g., to clarify, correct, suggest 
alternatives, etc) before the entire content is 
conveyed. This requires a rethinking of cur- 
rent practice in generation to make it incre- 
mental and interactive. 
• True interruptions - people may interrupt the 
system while it is talking. It is unclear at this 
stage what the system should assume was 
conveyed. The strategies of assuming noth- 
ing was conveyed, or that all was conveyed 
have obvious faults. We are pursuing alter- 
natives based on knowing when speech was 
interrupted, but using this ififormation suc- 
cessfully remains a difficult problem. 

References 
Allen, James et al, An Architecture for a Generic 
Dialogue Shell, to appear, J. Natural Language 
Engineering, 2000. 
Ferguson, George and J. Allen,-TRIPS: An Integrated 
Intelligent Problem-Solving Assistant, Proc. Na- 
tional Conference on AI (AAAI-98), Madison, WI, 
1998. 
