Cecile L. Paris

PLANNING TEXT FOR ADVISORY DIALOGUES

Explanation is an interactive process requiring a dialogue between advice-giver and advice-seeker. In this paper, we argue that in order to participate in a dialogue with its users, a generation system must be capable of reasoning about its own utterances and therefore must maintain a rich representation of the responses it produces. We present a text planner that constructs a detailed text plan, containing the intentional, attentional, and rhetorical structures of the text it generates.

Explanations in knowledge systems: design for explainable expert systems

The explainable expert systems framework (EES), in which the focus is on capturing those design aspects that are important for producing good explanations, including justifications of the systems actions, explications of general problem-solving strategies, and descriptions of the systems terminology, is discussed. EES was developed as part of the Strategic Computing Initiative of the US Dept. of Defenses Defense Advanced Research Projects Agency (DARPA). both the general principles from which the system was derived and how the system was derived from those principles can be represented in EES. The Program Enhancement Advisor, which is the main prototype on which the explanation work has been developed and tested, is presented. PEA is an advice system that helps users improve their Common Lisp programs by recommending transformations that enhance the users code. How EES produces better explanations is shown.<<ETX>>

Employing Knowledge Resources in a New Text Planner Architecture

We describe in this paper a new text planner that has been designed to address several problems we had encountered in previous systems. Motivating factors include a clearer and more explicit separation of the declarative and procedural knowledge used in a text generation system as well as the identification of the distinct types of knowledge necessary to generate coherent discourse, such as communicative goals, text types, schemas, discourse structure relations, and theme development patterns. This knowledge is encoded as separate resources and integrated under a flexible planning process that draws from appropriate resources whatever knowledge is needed to construct a text. We describe the resources and the planning process and illustrate the ideas with an example.

Exploiting user feedback to compensate for the unreliability of user models

Natural Language is a powerful medium for interacting with users, and sophisticated computer systems using natural language are becoming more prevalent. Just as human speakers show an essential, inbuilt responsiveness to their hearers, computer systems must “tailor” their utterances to users. Recognizing this, researchers devised user models and strategies for exploiting them in order to enable systems to produce the “best” answer for a particular user.Because these efforts were largely devoted to investigating how a user model could be exploited to produce better responses, systems employing them typically assumed that a detailed and correct model of the user was available a priori, and that the information needed to generate appropriate responses was included in that model. However, in practice, the completeness and accuracy of a user model cannot be guaranteed. Thus, unless systems can compensate for incorrect or incomplete user models, the impracticality of building user models will prevent much of the work on tailoring from being successfully applied in real systems. In this paper, we argue that one way for a system to compensate for an unreliable user model is to be able to react to feedback from users about the suitability of the texts it produces. We also discuss how such a capability can actually alleviate some of the burden now placed on user modeling. Finally, we present a text generation system that employs whatever information is available in its user model in an attempt to produce satisfactory texts, but is also capable of responding to the users follow-up questions about the texts it produces.

The role of the user's domain knowledge in generation

A question‐answering program that provides access to a large amount of data will be most useful if it can tailor its answers to each individual user. In particular, a users level of knowledge about the domain of discourse is an important factor in this tailoring if the answer provided is to be both informative and understandable to the user. In this research, we address the issue of how the users domain knowledge, or the level of expertise, might affect an answer. We present TAILOR, a flexible computer system that takes into account this knowledge to provide an answer that is appropriate for users with varying levels of expertise (including novices and experts), without requiring an a priori set of user types.

Requirements for an expert system explanation facility

For the past several years, we have worked on building an explanation component for an expert system building framework (or “shell”), the Explainable Expert System (EESQ Framework. In this short paper, we describe the characteristics that we believe to be essential for an explanation component of an expert system. We then identify important features of the EES architecture that support the desired capabilities. Finally, we discuss some areas where fruitful work remains to be done.

EXPECT: Intelligent Support for Knowledge Base Refinement

Effective knowledge acquisition amounts to having good sources of expectations that can provide guidance about what knowledge needs to be acquired from users. Current approaches to knowledge acquisition often rely on strong models of the problem-solving method used in the task domain to form expectations. These methods are often implicit in the tool, which is a strong limitation for their use in different domains. Additionally, these tools require an understanding of the method to be used that most experts find difficult to overcome. In this paper we present EXPECT, a novel approach to knowledge acquisition based on the EES architecture that forms expectations based on the current knowledge contained in the system about the task, and are not hard-coded in the tool. We show how the explicit representation of domain principles and its relation to compiled procedural knowledge enables a system to form expectations as to what knowledge is missing or incorrect. This capability coupled with a dialogue-based explanation facility makes communication with the knowledge acquisition tool more natural to domain experts.

Flexible Generation: Taking the User into Account

Sophisticated computer systems capable of interacting with people using natural language are becoming increasingly common. These systems need to interact with a wide variety of users in different situations. Typically, these systems have access to large amounts of data and must select from these data the information to present to the user. No single generated text will be adequate across all user types and all situations. Certainly, people plan what they will say or write based in part on their knowledge of the listener or intended reader. Similarly, computer systems that produce language must take their listeners/readers into account in order to be effective. In particular, the user’s level of knowledge about the domain of discourse is an important factor in this tailoring, if the text provided is to be both informative and understandable to the user. The text should not contain information that is already known or can be easily inferred, nor should it include facts that the user cannot understand. This paper demonstrates the feasibility of incorporating the user’s domain knowledge or user’s expertise, into a text generation system and addresses the issue of how this factor might affect the content, organization and phrasing of a text. We look at two applications domains: (i) generating descriptions of complex physical objects, and (ii) generating documentation for programming languages. We show how a computer generation system can make use of both stereotypical and individualized user models.