Exploiting user feedback to compensate for the unreliability of user models

Abstract

Abstract. Natural Language is apowerful 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. and her bachelor's degree from the University of California in Berkeley. Her research interests include natural language generation and user modeling, discourse, expert system explanation, human-computer interaction, intelligent tutoring systems, machine learning, and knowledge acquisition. At Columbia University, she developed a natural language generation system capable of producing multi-sentential texts tailored to the users" level of expertise about the domain. At ISI, she has been involved in designing a flexible explanation facility that supports dialogue for an expert system shell. 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. F~inally, 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 user's follow-up questions about the texts it produces.

Refining instructional text generation after evaluation

“…The first release of OPADE's text generator follows the line of several previous systems (among others [24,34,37]). Its most relevant feature is that the considered application field is not a 'toy domain'.…”

“…Nevertheless, a similar knowledge representation is common to the majority of discourse planners (see among others [1,7,34,37]), with which OPADE shares the same problems: 1. main effects of elementary and complex communicative acts are not represented adequately: negative as well as positive effects should be mentioned and should be described in a quantitative way (degree of persuasion and knowledge and so on). 2. side effects of acts are not mentioned; for example: the Describe drug administration increases the IUs' degree of knowledge about how to perform the action, but may affect, at the same time, their intention to perform the action; 3. effects produced by the communicative acts are not correlated to the situation; for example: the Request produces, in the Hearer, a lower increase of the degree of persuasion to perform the action if H is 'hostile' towards S, if the action is 'unpleasant' and so on.…”

Personalised hypermedia presentation techniques for improving online customer relationships

“…The stereotype approach has been used in many systems (Ambrosini et al, 1997;Fink et al, 1998;Kobsa et al, 1994;Moore & Paris, 1992, to name but a few). The importance of the stereotype approach is reflected by the fact that stereotyping is one of the few common features of the so-called "user modelling shell systems" that were built in the mid-1990s as toolboxes for the development of user modelling systems.…”

Untitled

“…In addition, Quilici's work focused primarily on understanding a user's feedback, while previous work on combining explanation planning and user feedback focused on generating a response (e.g. Moore and Paris, 1992and Cawsey, 1990 in Section 2.4). As Quilici himself noted, work in explanation and feedback could bene¢t by integrating these approaches.…”

“…These systems used users' responses to gather user modeling information required to plan their presentation. * Two of these systems focused on planning the content of their explanations (Moore and Paris, 1992;Peter and Ro« sner, 1994), while the remaining systems planned both the content and the type of their contribution (Cawsey, 1990(Cawsey, , 1993Shifroni and Shanon, 1992;Asami et al, 1996). Moore and Paris (1992) adopted the opposite approach to that of the systems described above, rejecting the reliance of NLG systems on complete and accurate user models, and arguing that a system's ability to react to a user's feedback compensates for the lack of reliability of user models.…”