Question‐Answering for Intelligent On‐Line Help: The Process of Intelligent Responding

Abstract

This article examines the process of specifying a question‐answering help facility in the context of UNIX mail. The specification was based upon experimental expert‐user facilitative dialogues. These dialogues were analyzed using a classification scheme developed for the purpose. The scheme provides a metalanguage for describing patterns of intent and rhetorical structure in dialogue. Using the scheme as a tool, common patterns in expert‐user dialogue emerged, providing insights into both tutoring strategy and the linguistic forms required to generate help output. These were refined to form a set of frames consisting of stereotypical patterns of rhetorical predicates to answer different types of questions which could be adapted to the needs of a particular user. The significance of the system developed from this specification is discussed.

Aides en ligne à l’utilisation de logiciels grand public : problèmes spécifiques de conception et solutions potentielles

“…Tours de parole et affichages sont datés et présentés dans l'ordre de leur occurrence (Falzon et Karsenty, 1997). Les interventions orales des experts dans les deux situations à contexte limité (soit dix dialogues, cinq par situation) ont été décomposées en actes de langage dont le contenu informationnel a été caractérisé à l'aide des taxonomies existantes, notamment (Pilkington, 1992). La méthodologie est détaillée dans (Capobianco, 2002).…”

“…Tableau 2 : Répartition des demandes d'informations des novices en fonction d'une taxonomie inspirée de (Pilkington, 1992). (Capobianco, 2002) "Conditions A et B", "E", "N" : voir la légende du tableau 1.…”

“…Cette voie de recherche a été considérée et abordée très tôt, mais les efforts engagés se sont avérés infructueux. Des tentatives dont l'objectif à long terme était la mise en oeuvre de dialogues d'aide en langue naturelle ont vu le jour dès le début des années 90 (Pilkington 1992 ;Roestler et McLellan, 1995). Mais elles n'ont pas été poursuivies, peut-être parce que les dialogues envisagés étaient complexes, et les échanges textuels.…”

Recognizing and Revising Unconvincing Explanations

“…SUSI (Ierrams-Smith, 1989) focused on providing 'over the shoulder' advice about command effects, but did not attempt to engage in dialogs with its users. UC (Wilensky et al, 1988) and PORSCHE (Pilkington, 1992;Tattersall, 1992) concentrated on understanding and answering questions about which commands to use to achieve a particular user goal, but did not attempt to dynamically address feedback about whether their answer was satisfactory. Similarly, the Correction Machine (Quilici, 1989a) and its earlier incarnations (Quilici, 1989b;Quilici et al, 1988) provided explanations for user misconceptions about UNIX, but did not attempt to update these explanations in response to user feedback.…”

“…Early research explored a variety of different explanation types and the necessary knowledge and techniques to produce them (Clancey, 1983(Clancey, , 1987Clancey and Letsinger, 1984;Hasling et al, 1984;Neches et al, 1985;Swartout, 1983). More recent work has focused on explanation planning and tailoring explanations to the user (Bateman and Paris, 1989;McKeown et al, 1985;Pilkington, 1992;Tattersall, 1992). Unfortunately, much less attention has been paid to the problem of explanation failure (Ringle and Bruce, 1981): what to do when an advice-seeking user is not satisfied with an advice-giving system's explanation.…”