U1, Speech in the interface:2. Dialogue Management1 Module u1: Speech in the Interface 2: Dialogue Management Jacques Terken HG room 2:40 tel. (247) 5254.

Slides:



Advertisements
Similar presentations
Facilitating Effective Meetings
Advertisements

Negotiative dialogue some definitions and ideas. Negotiation vs. acceptance Clark’s ladder: –1. A attends to B’s utterance –2. A percieves B’s utterance.
Supporting the IEP Process A Facilitator’s Guide Presentation adapted from: Martin, N. (2010). Supporting the IEP process: A facilitator’s guide. Baltimore,
SECOND MIDTERM REVIEW CS 580 Human Computer Interaction.
Clinical Supervision Foundations Module Six Performance Evaluation.
Fit to Learn Using the Employability Skills Framework to improve your performance at College The Employability Skills Framework has been developed by business.
Building & Leading Teams for Impact December 20, 2011.
From requirements to design
Error detection in spoken dialogue systems GSLT Dialogue Systems, 5p Gabriel Skantze TT Centrum för talteknologi.
SAI User-System Interaction U1, Speech in the interface: 6. Human communication1 Module u1: Speech in the Interface 6: Human Communication Jacques Terken.
CS CS 5150 Software Engineering Lecture 12 Usability 2.
Communication and interactive policy development Wednesday 2nd of July.
What can humans do when faced with ASR errors? Dan Bohus Dialogs on Dialogs Group, October 2003.
Part 4: Evaluation Chapter 20: Why evaluate? Chapter 21: Deciding on what to evaluate: the strategy Chapter 22: Planning who, what, where, and when Chapter.
Information, action and negotiation in dialogue systems Staffan Larsson Kings College, Jan 2001.
An evaluation framework
6/28/20151 Spoken Dialogue Systems: Human and Machine Julia Hirschberg CS 4706.
Lecture Nine Database Planning, Design, and Administration
Flexible User Interfaces for B2C Systems, Using Booking as a Case Study Kai A. Olsen Molde College, Norway Alessio Malizia, Dip. Scienze dell'Informazione,
U1, Speech in the interface: 1. Introduction1 Module u1: Speech in the Interface 1: Introduction Jacques Terken HG room 2:40 tel. (247) 5254
Towards Natural Clarification Questions in Dialogue Systems Svetlana Stoyanchev, Alex Liu, and Julia Hirschberg AISB 2014 Convention at Goldsmiths, University.
Jude Carroll, author of Tools for Teaching in an Educationally Mobile World (Routledge 2015) Supporting teaching across cultures: the role of good practice.
Dorothy Chun UC Santa Barbara.  Why do it? 1.To be able to say with greater certainty, beyond anecdotal evidence, that your efforts are having an effect.
22 November 2004 Multisession management in spoken dialogue system Hoá NGUYEN & Jean CAELEN.
Cross-cultural Communication and Negotiation
Innovative Schools toolkit
Browaeys and Price, Understanding Cross-cultural Management, 1 st Edition, © Pearson Education Limited 2009 Slide 7.1 Part Two: Understanding Cross-cultural.
DQA meeting: : Learning more effective dialogue strategies using limited dialogue move features Matthew Frampton & Oliver Lemon, Coling/ACL-2006.
Interactive Dialogue Systems Professor Diane Litman Computer Science Department & Learning Research and Development Center University of Pittsburgh Pittsburgh,
1 REQUIREMENT ENGINEERING Chapter 7. 2 REQUIREMENT ENGINEERING Definition Establishing what the customer requires from a software system. OR It helps.
Design Science Method By Temtim Assefa.
Working group on multimodal meaning representation Dagstuhl workshop, Oct
Interaction Design Session 12 LBSC 790 / INFM 718B Building the Human-Computer Interface.
Spoken dialog for e-learning supported by domain ontologies Dario Bianchi, Monica Mordonini and Agostino Poggi Dipartimento di Ingegneria dell’Informazione.
Speech and Language Processing Chapter 24 of SLP (part 3) Dialogue and Conversational Agents.
Eurostat Overall design. Presented by Eva Elvers Statistics Sweden.
Action Research for Teaching a Second Language By: Sarah Spivak.
Towards multimodal meaning representation Harry Bunt & Laurent Romary LREC Workshop on standards for language resources Las Palmas, May 2002.
Management Skills Different management styles draw more on some skills than others. For instance, - What style might managers with good people skills.
Module Nine: Emotional Communication (Conversation) 8- 1.
Dept. of Computer Science University of Rochester Rochester, NY By: James F. Allen, Donna K. Byron, Myroslava Dzikovska George Ferguson, Lucian Galescu,
1 Boostrapping language models for dialogue systems Karl Weilhammer, Matthew N Stuttle, Steve Young Presenter: Hsuan-Sheng Chiu.
Issues in Multiparty Dialogues Ronak Patel. Current Trend  Only two-party case (a person and a Dialog system  Multi party (more than two persons Ex.
16.0 Spoken Dialogues References: , Chapter 17 of Huang 2. “Conversational Interfaces: Advances and Challenges”, Proceedings of the IEEE,
1 Natural Language Processing Lecture Notes 14 Chapter 19.
PROGRAM THEORY What is it? Why is it important?. Most familiar form – Logic Model Where are you going? How will you get there? What will tell you that.
Information state and dialogue management in the TRINDI Dialogue Move Engine Toolkit, Larsson and Traum 2000 D&QA Reading Group, Feb 20 th 2007 Genevieve.
Techniques for Highly Effective Communication Professional Year Program - Unit 5: Workplace media and communication channels.
By Germaine Cheung Hong Kong Computer Institute
1 Knowledge Acquisition and Learning by Experience – The Role of Case-Specific Knowledge Knowledge modeling and acquisition Learning by experience Framework.
Welcome Back, Folks! We’re travelling to a littele bit far-end of Language in Use Studies EAA remains your faithful companion.
HITIQA: Scenario Based Question Answering Tomek Strzalkowski, et al The State University of New York at Albany Paul Kantor, et al Rutgers University Boris.
Independent Enquirers Learners process and evaluate information in their investigations, planning what to do and how to go about it. They take informed.
Introduction to Communicative Language Teaching Zhang Lu.
Evaluating learning gain in a SAC: Case studies of six low proficiency students Presenter: Ellie Law HASALD Presentation 2 Dec 2010.
Integrating Multiple Knowledge Sources For Improved Speech Understanding Sherif Abdou, Michael Scordilis Department of Electrical and Computer Engineering,
Winter 2007SEG2101 Chapter 121 Chapter 12 Verification and Validation.
Speech Processing 1 Introduction Waldemar Skoberla phone: fax: WWW:
Dialogue Modeling 2. Indirect Requests “Can I have a cup of coffee?”  One approach to dealing with these kinds of requests is by plan-based inference.
Grounding and Repair Joe Tepperman CS 599 – Dialogue Modeling Fall 2005.
Agent-Based Dialogue Management Discourse & Dialogue CMSC November 10, 2006.
WP6 Emotion in Interaction Embodied Conversational Agents WP6 core task: describe an interactive ECA system with capabilities beyond those of present day.
Week 2: Interviews. Definition and Types  What is an interview? Conversation with a purpose  Types of interviews 1. Unstructured 2. Structured 3. Focus.
Predicting and Adapting to Poor Speech Recognition in a Spoken Dialogue System Diane J. Litman AT&T Labs -- Research
COMMUNICATION OF MEANING
GIVING FEEDBACK ON PERFORMANCE CONCERNS IN A 1:1 MEETING -
As You Enter Take a moment to network and exchange contact information from those in the room you do not have yet.
Integrating Learning of Dialog Strategies and Semantic Parsing
Managing Dialogue Julia Hirschberg CS /28/2018.
GIVING FEEDBACK ON PERFORMANCE CONCERNS IN A 1:1 MEETING -
Presentation transcript:

U1, Speech in the interface:2. Dialogue Management1 Module u1: Speech in the Interface 2: Dialogue Management Jacques Terken HG room 2:40 tel. (247) 5254

U1, Speech in the interface:2. Dialogue Management2 contents 1. Tasks of the dialogue manager 2. Initiative/Control 3. Dialogue structure 4. Dealing with channel and technology limitations

U1, Speech in the interface:2. Dialogue Management3 Dialogue phenomena n Natural language phenomena: User: “a flight from Boston to New York” System: “7:15 with Continental” User: “A later one?” [ flight from B. to NY.]

U1, Speech in the interface:2. Dialogue Management4 n Recognition of user’s dialogue act U: “I want a flight from Boston to New York” syntactically: statement pragmatically: request for information n Dealing with recognition errors, misunderstandings and other communication problems U: A flight from Boston to New York … S; I’ve booked a flight from Houston to Newark … In addition, users often have problems to know what kind of reaction is expected for successful communication

U1, Speech in the interface:2. Dialogue Management5 DM tasks n Interpreting user contribution in dialogue context and situational context –Dealing with natural language phenomena –Recognition of user’s dialogue act –Involves dealing with recognition errors, misunderstandings and other communication problems n Deciding on next system contribution

U1, Speech in the interface:2. Dialogue Management6 Deciding upon next system contribution n Interaction as co-operation –User and system co-operate to achieve a goal –transmission of information across communication channel may induce distortions  distinction between task-oriented acts and dialogue control acts 

U1, Speech in the interface:2. Dialogue Management7 n Task-oriented dialogue acts –Bring the dialogue purpose closer –WH-question, YN-question, Inform, WH-answer, … n Dialogue control acts –Directed towards keeping dialogue on the track and prevent and deal with problems, e.g. Requests for clarification Verification Feedback/confirmation Error recovery strategies But also greetings, apologies etc.

U1, Speech in the interface:2. Dialogue Management8 contents 1. Tasks of the dialogue manager 2. Initiative/Control 3. Dialogue structure 4. Dealing with channel and technology limitations

U1, Speech in the interface:2. Dialogue Management9 Initiative n System initiative S: Where are you travelling to U: London S: What day do you wish to travel U: Friday S: At what time U: 9 a.m

U1, Speech in the interface:2. Dialogue Management10 n User initiative: U: How many employees living in the London area earn more than ₤ S: Fifty four U: How many are female S: Eleven U: And managers S: Nine

U1, Speech in the interface:2. Dialogue Management11 n Mixed initiative S: Where are you travelling to U: I want to fly to London on Friday S: At what time do you want to fly to London U: Are there any cheap flights

U1, Speech in the interface:2. Dialogue Management12 contents 1. Tasks of the dialogue manager 2. Initiative/Control 3. Dialogue structure 4. Dealing with channel and technology limitations

U1, Speech in the interface:2. Dialogue Management13 Finite-state dialogue grammar n Dialogue structure represented as finite state transition network Departure town  arrival town  date  time  carrier n With repair sub-dialogues: abcdabcd rrrr

U1, Speech in the interface:2. Dialogue Management14 Destination? Was that $Destination? Day? Was that $Day? yesno yes London Friday

U1, Speech in the interface:2. Dialogue Management15 Frame-based approach n Dialogue Frame: Departure airport Destination airport Date Time Carrier n Onset condition: All parameters unknown n Condition action pairs produce dialogue structure: Condition: Origin and destination unknown Question: Which route do you want to travel

U1, Speech in the interface:2. Dialogue Management16 DestinationDayTime

U1, Speech in the interface:2. Dialogue Management17 Constraint relaxation and narrowing n “I want a flight from Boston to New York Friday morning, leaving between 7:15 and 7:45” n If no records are found, ask user to be more general n If many options are found (e.g. ≥ 5), ask user to be more specific n If 1 ≤ number of found options < N (e.g. 5), generate response that outputs the retrieved records: “I found the following flights: ….”

U1, Speech in the interface:2. Dialogue Management18 Cooperative dialogue n Don’t ask user to be more specific or more general, but propose solutions E.g. U: Is there a flight from Boston to New York between 7:15 and 7:45? S’: No S’’: No, but there is a flight at 7:55. Do you want me to book that one? n Requires understanding the user’s intentions and priorities: S’’’: * No but there is a flight to Philadelphia between 7:15 and 7:45

U1, Speech in the interface:2. Dialogue Management19 Theoretical frameworks for dialogue management n Plan-based approaches Dialogue as a special instance of rational (goal-directed) behaviour System job is to discover and react adequately to the speaker’s plan rather than the utterance n Modelling of Beliefs-Desires-Intentions Adds modeling of assumptions and beliefs that speaker has, in order to identify common ground n Rational agency (dialogue management from first principles) n Agent-based approaches (combination of plan-based approaches and local repair mechanisms with heterogeneous control)

U1, Speech in the interface:2. Dialogue Management20 contents 1. Tasks of the dialogue manager 2. Initiative/Control 3. Dialogue structure 4. Dealing with channel and technology limitations

U1, Speech in the interface:2. Dialogue Management21 n Technology limitations: speech recognition errors n Channel limitations: –Speech is sequential and volatile n Requires feedback at different levels of communication: –Perception –Interpretation –Evaluation  Grounding theory (Clark): dialogue as joint activity; reaching agreement over what was said and meant

U1, Speech in the interface:2. Dialogue Management22 Dialogue strategies n Zooming Asking general questions first and narrowing down if no adequate response is received “How may I help you” n Adequate prompts are important –They inform the user about what kind of reply is expected –Require iterative design in combination with user studies (collecting extensive set of user reactions)

U1, Speech in the interface:2. Dialogue Management23 Verification strategies n Verification provides feedback and ensures that both participants stay in agreement about the state of the dialogue (what was said and agreed) n Explicit verification –So you want to go to New York n Implicit verification –What time do you want to arrive in New York n Confidence-based verification –Repeat question with very low confidence –Explicit verification with low to intermediate confidence –Implicit verification with higher confidence –No verification with very high confidence

U1, Speech in the interface:2. Dialogue Management24 Project n Exercises –Start from the original pizza application –Implement different verification strategies, both explicit and implicit –Try the system while simulating speech recognition errors Hint: wreck one of the options and run a scenario n Project –Conduct a task analysis (or construct use cases) and define the basic dialogue structure