We have been examining mixed-initiative planning systems in the context of command and control or logistical overview situations. In such environments, the human and the computer must work together in a very tightly coupled way to solve problems that neither alone could manage. In this paper, we describe our implementation of a prototype version of such a system, TRAINS-95, which helps a manager solve routing problems in a simple transportation domain. Interestingly perhaps, traditional planning technology does not play a major role in the system, and in fact it is difficult to see how such components might fit into a mixed-initiative system. We describe some of these issues, and present our agenda for future research into mixed-initiative plan reasoning. At this writing, the TRAINS-95 system has been used by more than 100 people to solve simple problems at various conferences and workshops, and in our experiments.

Natural language generation (NLG) is first and foremost a reasoning task. In this reasoning, a system plans a communicative act that will signal key facts about the domain to the hearer. In generating action descriptions, this reasoning draws on characterizations both of the causal properties of the domain and the states of knowledge of the participants in the conversation. This dissertation shows how such characterizations can be specified declaratively and accessed efficiently in NLG. The heart of this dissertation is a study of logical statements about knowledge and action in modal logic. By investigating the proof-theory of modal logic from a logic programming point of view, I show how many kinds of modal statements can be seen as straightforward instructions for computationally manageable search, just as Prolog clauses can. These modal statements provide sufficient expressive resources for an NLG system to represent the effects of actions in the world or to model an addressee whose knowledge in some respects exceeds and in other respects falls short of its own. To illustrate the use of such statements, I describe how the SPUD sentence planner exploits a modal knowledge base to

This paper describes a problem domain that lends itself to mixed initiative interaction. The domain is traded control with an autonomous robot. Traded control is a situation in which ahuman wants to control a robot during part of a task and the robot is autonomous during other parts of a task. A signi cant problem in traded control situations is that the robot doesn't know how the environment has been changed or what parts of the task have been accomplished when the human has been in control. Because of this, errors can occur when the human relinquishes control back to the robot � these errors can cause potentially dangerous situations. Our solution is to use an intelligent software architecture designed for autonomous robot control and modify it to work in concert with human control. Using an architecture designed for autonomy allows us to use the monitoring functions designed to track the actions of the robot to monitor the actions of human agents for the same tasks. The intelligent software architecture includes a mixed initiative planner, an execution monitor, robotic skills and a user interface. This paper describes the problem domain and our initial attempts at de ning a software architecture that operates in the domain.

We present a logic programming framework implemented over Prolog which is able to model an agent's mental state. An agent is modeled by a set of extended logic programming rules representing the agent's behavior, attitudes (beliefs, intentions, and goals), world knowledge, and temporal and reasoning procedures. At each stage the agents's mental state is defined by the well founded model of the extended logic program plus some constraints. Via this modeling an agent is able to interact with other agents, updating and revising its mental state after each event. The revision process includes the ability to remove contradictions in the agent's mental state. It is shown how this framework can handle interactions between agents with different behavior rules, namely, with different levels of cooperativeness and credulity. 1 Introduction In order to interact with other agents, an agent needs the ability to model its mental state. Namely, it is necessary to represent its attitudes (beliefs,...

We describe the goals, architecture, and functioning of the trains-93 system, with emphasis on the representational issues involved in putting together a complex language processing and reasoning agent. The system is intended as an experimental prototype of an intelligent, conversationally proficient planning advisor in a dynamic domain of cargo trains and factories. For this team effort, our strategy at the outset was to let the designers of the various language processing, discourse processing, plan reasoning, execution and monitoring modules choose whatever representations seemed best suited for their tasks, but with the constraint that all should strive for principled, general approaches. Disparities between modules were bridged by careful design of the interfaces, based on regular in-depth discussion of issues encountered by the participants. Because of the goal of generality and principled representation, the multiple representations ended up with a good deal in common (...

Designing an agent to participate in natural conversation requires more than just adapting a standard agent model to perceive and produce language. In particular, the model must be augmented with social attitudes (including mutual belief, shared plans, and obligations) and a notion of discourse context. The dialogue manager of the TRAINS-93 NL conversation system embodies such an augmented theory of agency. This paper focuses on the representation of mental state and discourse context and the deliberation strategies used in the agent model of the dialogue manager. 1 INTRODUCTION A dialogue manager is that part of a dialogue system that connects the I/O devices and translators (whether they be spoken or typed language, a command language, menu selection, graphical presentation, etc.) to the parts that do the domain task reasoning and performance. In a simple language front-end system (e.g., for querying a database), dialogue management can be little more than a transducer from the I/O ...

Abstract. Temporal reasoning started to be considered as a subject of study in artificial intelligence in the late ’70s. Since that several ways to represent and use temporal knowledge have been suggested. As a result of that there are several formalisms capable of coping with temporal notions in some way or other. They range from isolated proposals to complexsystems where the temporal aspect is used together with other important features for the task of modelling an intelligent agent. The purposes of this article are to summarize logic-based temporal reasoning research and give a glance on the different research tracks envisaging future lines of research. It is intended to be useful to those who need to be involved in systems having these characteristics and also an occasion to present newcomers some problems in the area that still wait for a solution.

The focus of this thesis proposal is to improve the ability of a computational system to understand spoken utterances in a dialogue with a human. Available computational methods for word recognition do not perform as well on spontaneous speech as we would hope. Even a state of the art recognizer achieves slightly worse than 70% word accuracy on (nearly) spontaneous speech in a conversation about a specific problem. To address this problem, I will explore novel methods for post-processing the output of a speech recognizer in order to correct errors. I adopt statistical techniques for modeling the noisy channel from the speaker to the listener in order to correct some of the errors introduced there. The statistical model accounts for frequent errors such as simple word/word confusions and short phrasal problems (one-to-many word substitutionsand many-to-one word concatenations). To use the model, a search algorithm is required to find the most likely correction of a given word sequence ...

This paper describes features of the active logic formalism as a tool for implementing mixed initiative intelligent systems. A framework is provided for assessing systems as "mixed-initiative", which is used to explore the relevant features both in the active logic formalism and associated implementations based upon and using that formalism. Active logics were developed as a means of combining the best of two worlds -- inference and reactivity -- without giving up much of either. This requires a special evolvingduring -inference model of time. Active logics are able to react to incoming information (including dialogue utterances by a collaborative partner) while reasoning is ongoing, blending new inputs into its inferences without having to start up a new theorem-proving effort. An implementation of active logic is described, which also includes special features for reasoning about and performing actions. This implementation is also used as the backbone of a dialogue system. 1 Introduc...

. So far, well-founded planning concentrates ultimately on the generation of action sequences which produce desired behaviors. But, to promote a planner's success in human assistance and support, varied representational and operational flexibility is needed. The paper argues that a compositional temporal logic framework is well-suited to reach the necessary flexibility. Planning is viewed as an inference process based on various refinements and plans work as transmitters of appropriate information about this process. Among other things nonlinear planning is explained in terms of concurrent refinements. 1 Introduction Whatever kind of planner is used, the ultimate aim of well-founded classical planning is to find a ground operator sequence, which when executed in the given initial state, will produce desired behaviors. Thereby, most techniques used have concentrated on the sub-class of behavioral constraints called the goals of attainment, i.e. a single goal state has been specified. S...