Curry I. Guinn

Mixed-initiative interaction

James F. Allen, University of Rochester Mixed-initiative interaction is a key aspect of effective human-computer interaction and has great potential to affect work on multiagent systems. The term mixed-initiative interactionis sometimes conflated with human-computer interaction itself, but this is a mistake because almost all models of HCI so far are not mixedinitiative, and mixed-initiative systems need not involve a human. It is perhaps in HCI where we will see the greatest impact. The development of mixed-initiative intelligent systems will ultimately revolutionize the world of computing even more than the recent move to GUIs, in my view. This essay describes the goals of research in mixed-initiative interaction, suggests a general framework for thinking about work in the area based on the properties of human dialogue, and then briefly describes the key problems to overcome before mixed-initiative systems become a reality. For simplicity, I will focus on a single scenario consisting of two agents: a human and an intelligent system. Mixed-initiative interaction can occur in many other scenarios as well, including between multiple machines cooperating to perform tasks (such as in distributed planning) or between multiple people and machines interacting to coordinate their activities (collaboration systems, for example). Most everything I say here generalizes to these other cases. In fact, many of the issues become even more crucial as the number of agents grows. In many examples, I will draw from our experience in building mixed-initiative planning systems over the last five years. 1–3.

Lessons learned in modeling schizophrenic and depressed responsive virtual humans for training

This paper describes lessons learned in developing the linguistic, cognitive, emotional, and gestural models underlying virtual human behavior in a training application designed to train civilian police officers how to recognize gestures and verbal cues indicating different forms of mental illness and how to verbally interact with the mentally ill. Schizophrenia, paranoia, and depression were all modeled for the application. For linguistics, the application has quite complex language grammars that captured a range of syntactic structures and semantic categories. For cognition, there is a great deal of augmentation to a plan-based transition network needed to model the virtual humans knowledge. For emotions and gestures, virtual human behavior is based on expert-validated mapping tables specific to each mental illness. The paper presents five areas demanding continued research to improve virtual human behavior for use in training applications

Mechanisms for Mixed-Initiative Human-Computer Collaborative Discourse

In this paper, we examine mechanisms for automatic dialogue initiative setting. We show how to incorporate initiative changing in a task-oriented human-computer dialogue system, and we evaluate the effects of initiative both analytically and via computer-computer dialogue simulation.

An Analysis of Initiative Selection in Collaborative Task-Oriented Discourse

In this paper we propose a number of principles and conjectures for mixed-initiative collaborative dialogs. We explore some methodologies for managing initiative between conversational participants. We mathematically analyze specific initiative-changing mechanisms based on a probabilistic knowledge base and user model. We look at the role of negotiation in managing initiative and quantify how the negotiation process is useful toward modifying user models. Some experimental results using computer–computer simulations are presented along with some discussion of how such studies are useful toward building human–computer systems.

An Analysis of Initiative Selection in CollaborativeTask-Oriented Discourse

In this paper we propose a number of principles and conjectures for mixed-initiative collaborative dialogs. We explore some methodologies for managing initiative between conversational participants. We mathematically analyze specific initiative-changing mechanisms based on a probabilistic knowledge base and user model. We look at the role of negotiation in managing initiative and quantify how the negotiation process is useful toward modifying user models. Some experimental results using computer–computer simulations are presented along with some discussion of how such studies are useful toward building human–computer systems.

Efficient collaborative discourse: a theory and its implementation

An architecture for voice dialogue machines is described with emphasis on the problem solving and high level decision making mechanisms. The architecture provides facilities for generating voice interactions aimed at cooperative human-machine problem solving. It assumes that the dialogue will consist of a series of local self-consistent subdialogues each aimed at subgoals related to the overall task. The discourse may consist of a set of such subdialogues with jumps from one subdialogue to the other in a search for a successful conclusion. The architecture maintains a user model to assure that interactions properly account for the level of competence of the user, and it includes an ability for the machine to take the initiative or yield the initiative to the user. It uses expectation from the dialogue processor to aid in the correction of errors from the speech recognizer.

An evaluation of virtual human technology in informational kiosks

In this paper, we look at the results of using spoken language interactive virtual characters in information kiosks. Users interact with synthetic spokespeople using spoken natural language dialogue. The virtual characters respond with spoken language, body and facial gesture, and graphical images on the screen. We present findings from studies of three different information kiosk applications. As we developed successive kiosks, we applied lessons learned from previous kiosks to improve system performance. For each setting, we briefly describe the application, the participants, and the results, with specific focus on how we increased user participation and improved informational throughput. We tie the results together in a lessons learned section.

Goal-Oriented Multimedia Dialogue with Variable Initiative

A dialogue algorithm is described that executes Prolog-style rules in an attempt to achieve a goal. The algorithm selects paths in the proof in an attempt to achieve success and proves subgoals on the basis of internally available information where possible. Where “missing axioms” are discovered, the algorithm interacts with the user to solve subgoals, and it uses the received information from the user to attempt to complete the proof. A multimedia grammar codes messages sent to and received from the user into a combination of speech, text, and voice tokens. This theory is the result of a series of dialog projects implemented in our laboratory. These will be described including statistics that measure their levels of success.

Teaching a hierarchical model of computation with animation software in the first course

In a world saturated with computers, it is important that the popnlace have some understanding of what thesedevices am, how they work, what they can do, and what they cannot do. People will not intelligently live with machines if they regard them as indistinguishable from magic. In fact, many authors (Brookshear(1988), Schaffer(1988), Tucker (1992)) have called for &eased breadth in the early computer science courses specifically to better educate nonmajors who may never take another computer course and to give perspective to majors who are looking forward to a series of specialized courses. In the education of majors, many recent committees and writers haveargued for kreased breadth in the early courses including Denning et al. (1989), the ACIWIEEE-CS Joint Curriculum Task Force (1991), and Barker et al (1992).

A comparison of syntax, semantics, and pragmatics in spoken language among residents with Alzheimer's disease in managed-care facilities

This research is a discriminative analysis of conversational dialogues involving individuals suffering from dementia of Alzheimers type. Several metric analyses are applied to the transcripts of the Carolina Conversation Corpus in order to determine if there are significant statistical differences between individuals with and without Alzheimers disease. Our prior research suggests that there exist measurable linguistic differences between managed-care residents diagnosed with Alzheimers disease and their caregivers. This paper presents results comparing managed-care residents diagnosed with Alzheimers disease to other managed-care residents. Results from the analysis indicate that part-of-speech and lexical richness statistics may not be good distinguishing attributes. However, go-ahead utterances and certain fluency measures provide defensible means of differentiating the linguistic characteristics of spontaneous speech between individuals that are and are not diagnosed with Alzheimers disease. Two machine learning algorithms were able to classify the speech of individuals with and without dementia of the Alzheimers type with accuracy up to 80%.