Inhalt

•  
   Abstract
•  
   Acknowledgements
•  
  Preliminary Remarks
•  
   Speakers and listeners
•  
   Previously published material and co-authorship
•  
  1 Introduction
•  
   1.1 Context and motivation
•  
   1.2 Objective and scope
•  
   Listener state attribution
•  
   Interactive adaptation
•  
   Feedback elicitation
•  
   1.3 Overview
•  
   I Dialogue Coordination in the Human and the Machine
•  
  2 Communication, Dialogue, and Coordination
•  
   2.1 Understanding, misunderstanding, and non-understanding
•  
   2.1.1 Understanding
•  
   2.1.2 Strong and weak concepts of understanding
•  
   2.1.3 Non-understanding and misunderstanding
•  
   2.1.4 Ways out: Approaching understanding
•  
   2.2 Common ground and grounding
•  
   2.2.1 Grounding
•  
   2.2.2 Computational theories of grounding
•  
   2.3 Alignment, adaptation, and coordination
•  
   2.3.1 Interactive adaptation
•  
   2.3.2 Interactive alignment
•  
   2.3.3 Full common ground
•  
   2.3.4 Monitoring and adjustment
•  
   2.3.5 Minimal partner models
•  
   2.3.6 Intermediate summary
•  
   2.4 Reaching understanding with artificial conversational agents
•  
   2.5 Summary and conclusion
•  
  3 Communicative Feedback
•  
   3.1 On the origins of the concept of feedback in communication
•  
   3.2 Terminology
•  
   3.3 Form and structure of feedback signals
•  
   3.3.1 Short verbal/vocal feedback
•  
   3.3.2 Prosody and intonation of feedback
•  
   3.3.3 Embodied, non-verbal feedback
•  
   3.3.4 Intermediate summary: form
•  
   3.4 Meaning and function of listener feedback
•  
   3.4.1 Hierarchical relations
•  
   3.4.2 Polarity
•  
   3.4.3 Awareness and intentionality
•  
   3.5 Placement and timing
•  
   3.6 Communicative feedback in artificial conversational agents
•  
   3.7 Summary and conclusion
•  
   II A Computational Model of Attentive Speaking
•  
   4 Interactional Intelligence for Attentive Speaking
•  
  5 Mental State Attribution Based on Communicative Listener Feedback
•  
   5.1 Introduction
•  
   5.2 A causal model of the interaction
•  
   5.3 The attributed listener state (ALS)
•  
   5.4 Feedback and attributed listener state
•  
   5.5 Interpreting listener feedback in context
•  
   5.6 Feedback and grounding
•  
   5.7 Interpreting listener feedback in an evolving context
•  
   5.7.1 Dynamic minimal mentalising
•  
   5.7.2 Worked example
•  
   5.7.3 Discourse structure and belief state evolution
•  
   5.8 Related work
•  
   5.9 Discussion
•  
   5.10 Summary
•  
  6 Interactive Adaptation
•  
   6.1 Levels and mechanisms of adaptation
•  
   6.2 Adaptive natural language generation
•  
   6.2.1 The SPUD microplanning framework
•  
   6.2.2 Incremental generation with SPUD_inc
•  
   Generation example
•  
   Discussion
•  
   6.2.3 Adaptive Generation in SPUD_ia
•  
   6.2.4 Adaptation mechanisms in SPUD_ia
•  
   Verbosity
•  
   Redundancy
•  
   6.3 Adaptive incremental speech synthesis and behaviour realisation
•  
   6.4 Summary and discussion
•  
  7 Feedback Elicitation
•  
   7.1 Seeking evidence of understanding
•  
   7.2 Information needs and feedback requests
•  
   7.3 Criteria for eliciting feedback
•  
   7.4 Eliciting feedback
•  
   7.5 Summary and conclusion
•  
   III Evaluation
•  
  8 Bringing it Together: An Attentive Speaker Agent
•  
   8.1 Overall model and architecture
•  
  8.2 Incremental processing
•  
   8.2.1 Incremental processing for artificial conversational agents
•  
   8.2.2 The IU-model
•  
  8.3 Behaviour planning and realisation
•  
   8.3.1 The SAIBA-architecture for behaviour generation
•  
   8.3.2 Real-time generation of multimodal behaviour
•  
   8.3.3 Behaviour generation for attentive speaking
•  
   8.3.4 Dialogue Engine
•  
   8.3.5 Natural language generation with SPUD_ia
•  
   8.3.6 Gaze behaviour
•  
  8.4 Attributed Listener State
•  
   8.4.1 Collecting evidence
•  
   8.4.2 Processing evidence
•  
   8.4.3 A note recognising listener feedback signals
•  
   8.5 Personal calendar assistant scenario
•  
   8.6 Summary
•  
  9 Evaluation of the Attentive Speaker Agent
•  
   9.1 Evaluating artificial conversational agents
•  
   9.2 Evaluation strategy and general hypotheses
•  
  9.3 Materials and methods
•  
   9.3.1 Objective metrics and variables
•  
   9.3.2 Subjective metrics and variables
•  
   9.3.3 Experimental conditions
•  
   9.3.4 Setup
•  
   9.3.5 The Wizard-of-Oz
•  
   9.3.6 Procedure
•  
   9.4 Participants
•  
  9.5 Analysis and results
•  
  9.5.1 Participants' feedback behaviour
•  
   Annotation of participant's feedback signal
•  
   Do participants provide natural feedback in human–agent interaction?
•  
   Does the agent's behaviour influence participants' feedback rate?
•  
   Does participants' feedback rate change over time?
•  
   Do participants (just) respond to feedback cues?
•  
   Intermediate summary
•  
   9.5.2 Objective quality of the interaction
•  
   Analysing costs: interaction durations
•  
   Analysing costs: repetitions
•  
   Intermediate summary: Costs
•  
   Analysing performance: understanding in terms of recall
•  
   Intermediate summary: Performance
•  
   Efficiency: Recall score versus duration or repetition
•  
   Intermediate summary: Efficiency
•  
   9.5.3 Subjective quality of the interaction
•  
  The agent's communicative competence (Q1–Q3)
•  
   Q1
•  
   Q2
•  
   Q3
•  
   The agent's attentive speaking capabilities (Q4–Q10)
•  
   Q4
•  
   Q5
•  
   Q6
•  
   Q7
•  
   Q8
•  
   Q9
•  
   Q10
•  
  The agent's helpfulness (Q11–Q14)
•  
   Q11
•  
   Q12
•  
   Q13
•  
   Q14
•  
   The agent's naturalness (Q15–Q17)
•  
   Q15
•  
   Q16
•  
   Q17
•  
  Perception of the task and study
•  
   Q18
•  
   Q19
•  
   Q20
•  
   Intermediate summary
•  
   9.6 General discussion
•  
   Overall conclusion
•  
   Future work
•  
  10 Conclusion
•  
   10.1 Summary
•  
  10.2 Contributions and implications
•  
   Communicative listener feedback
•  
   Incremental processing in conversational agents
•  
   Grounding and interactive adaptation
•  
   Intelligent virtual agents
•  
   Interactional intelligence and dialogue coordination
•  
   10.3 Limitations and future research directions
•  
   Appendices
•  
  A Model Parametrisation from Implicit Representation
•  
   A.1 Model parametrisation
•  
   A.2 The CPT generation algorithm
•  
   A.3 Example – Generating a CPT for Pr(U | P, FB)
•  
  B Study Materials
•  
   B.1 Short written instructions
•  
   B.2 Detailed oral instructions
•  
   Bibliography
•  
   Accompanying Resources
•  
   Copyright Permissions
•  
   Affidavit