(Hidden) Information State Models Ling575 Discourse and Dialogue May 25, 2011.
Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.
-
date post
19-Dec-2015 -
Category
Documents
-
view
227 -
download
2
Transcript of Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.
![Page 1: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/1.jpg)
Dialogue Management
Ling575Discourse and Dialogue
May 18, 2011
![Page 2: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/2.jpg)
Dialog Management TypesFinite-State Dialog Management
Frame-based Dialog Management InitiativeVoiceXMLDesign and evaluation
Information State ManagementDialogue Acts
Recognition & generation
Statistical Dialogue Managemant (POMDPs)
![Page 3: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/3.jpg)
Finite-State Management
![Page 4: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/4.jpg)
Pros and ConsAdvantages
Straightforward to encodeClear mapping of interaction to modelWell-suited to simple information accessSystem initiative
DisadvantagesLimited flexibility of interaction
Constrained input – single itemFully system controlledRestrictive dialogue structure, order
Ill-suited to complex problem-solving
![Page 5: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/5.jpg)
Frame-based DialogueManagement
Finite-state too limited, stilted, irritating
More flexible dialogue
![Page 6: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/6.jpg)
Frame-based Dialogue Management
Essentially form-fillingUser can include any/all of the pieces of
formSystem must determine which entered,
remain
![Page 7: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/7.jpg)
Frame-based Dialogue Management
Essentially form-fillingUser can include any/all of the pieces of
formSystem must determine which entered,
remain
![Page 8: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/8.jpg)
Frame-based Dialogue Management
Essentially form-fillingUser can include any/all of the pieces of formSystem must determine which entered, remain
System may have multiple framesE.g. flights vs restrictions vs car vs hotelRules determine next action, question, information
presentation
![Page 9: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/9.jpg)
Frames and InitiativeMixed initiative systems:
A) User/System can shift control arbitrarily, any timeDifficult to achieve
![Page 10: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/10.jpg)
Frames and InitiativeMixed initiative systems:
A) User/System can shift control arbitrarily, any timeDifficult to achieve
B) Mix of control based on prompt type
Prompts:
![Page 11: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/11.jpg)
Frames and InitiativeMixed initiative systems:
A) User/System can shift control arbitrarily, any timeDifficult to achieve
B) Mix of control based on prompt type
Prompts:Open prompt:
![Page 12: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/12.jpg)
Frames and InitiativeMixed initiative systems:
A) User/System can shift control arbitrarily, any timeDifficult to achieve
B) Mix of control based on prompt type
Prompts:Open prompt: ‘How may I help you?’
Open-ended, user can respond in any wayDirective prompt:
![Page 13: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/13.jpg)
Frames and InitiativeMixed initiative systems:
A) User/System can shift control arbitrarily, any timeDifficult to achieve
B) Mix of control based on prompt type
Prompts:Open prompt: ‘How may I help you?’
Open-ended, user can respond in any wayDirective prompt: ‘Say yes to accept call, or no
o.w.’Stipulates user response type, form
![Page 14: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/14.jpg)
Initiative, Prompts, Grammar
Prompt type tied to active grammarSystem must recognize suitable input
Restrictive vs open-ended
![Page 15: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/15.jpg)
Initiative, Prompts, Grammar
Prompt type tied to active grammarSystem must recognize suitable input
Restrictive vs open-ended
Shift from restrictive to openTune to user: Novice vs Expert
![Page 16: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/16.jpg)
Initiative, Prompts, Grammar
Prompt type tied to active grammarSystem must recognize suitable input
Restrictive vs open-ended
Shift from restrictive to openTune to user: Novice vs Expert
![Page 17: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/17.jpg)
Dialogue Management:Confirmation
Miscommunication common in SDS“Error spirals” of sequential errors
Highly problematicRecognition, recovery crucial
Confirmation strategies can detect, mitigateExplicit confirmation:
![Page 18: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/18.jpg)
Dialogue Management:Confirmation
Miscommunication common in SDS“Error spirals” of sequential errors
Highly problematicRecognition, recovery crucial
Confirmation strategies can detect, mitigateExplicit confirmation:
Ask for verification of each input Implicit confirmation:
![Page 19: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/19.jpg)
Dialogue Management:Confirmation
Miscommunication common in SDS“Error spirals” of sequential errors
Highly problematicRecognition, recovery crucial
Confirmation strategies can detect, mitigateExplicit confirmation:
Ask for verification of each input Implicit confirmation:
Include input information in subsequent prompt
![Page 20: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/20.jpg)
Confirmation StrategiesExplicit:
![Page 21: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/21.jpg)
Confirmation Strategy Implicit:
![Page 22: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/22.jpg)
Pros and ConsGrounding of user input
Weakest grounding I.e. continued att’n, next relevant contibution
![Page 23: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/23.jpg)
Pros and ConsGrounding of user input
Weakest grounding insufficient I.e. continued att’n, next relevant contibution
Explicit:
![Page 24: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/24.jpg)
Pros and ConsGrounding of user input
Weakest grounding insufficient I.e. continued att’n, next relevant contibution
Explicit: highest: repetition Implicit:
![Page 25: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/25.jpg)
Pros and ConsGrounding of user input
Weakest grounding insufficient I.e. continued att’n, next relevant contibution
Explicit: highest: repetition Implicit: demonstration, display
Explicit;
![Page 26: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/26.jpg)
Pros and ConsGrounding of user input
Weakest grounding insufficient I.e. continued att’n, next relevant contibution
Explicit: highest: repetition Implicit: demonstration, display
Explicit;Pro: easier to correct; Con: verbose, awkward, non-
human
Implicit:
![Page 27: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/27.jpg)
Pros and ConsGrounding of user input
Weakest grounding insufficient I.e. continued att’n, next relevant contibution
Explicit: highest: repetition Implicit: demonstration, display
Explicit;Pro: easier to correct; Con: verbose, awkward, non-
human
Implicit:Pro: more natural, efficient; Con: less easy to correct
![Page 28: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/28.jpg)
RejectionSystem recognition confidence is too low
System needs to repromptOften repeatedly
![Page 29: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/29.jpg)
RejectionSystem recognition confidence is too low
System needs to repromptOften repeatedly
Out-of-vocabulary, out-of-grammar inputs
Strategies: Progressive prompting
![Page 30: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/30.jpg)
RejectionSystem recognition confidence is too low
System needs to repromptOften repeatedly
Out-of-vocabulary, out-of-grammar inputs
Strategies: Progressive prompting Initially: ‘rapid reprompting’: ‘What?’, ‘Sorry?’
![Page 31: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/31.jpg)
RejectionSystem recognition confidence is too low
System needs to repromptOften repeatedly
Out-of-vocabulary, out-of-grammar inputs
Strategies: Progressive prompting Initially: ‘rapid reprompting’: ‘What?’, ‘Sorry?’Later: increasing detail
![Page 32: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/32.jpg)
Progressive prompting
![Page 33: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/33.jpg)
VoiceXMLW3C standard for simple frame-based dialogues
Fairly common in commercial settings
Construct forms, menusForms get field data
Using attached promptsWith specified grammar (CFG)With simple semantic attachments
![Page 34: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/34.jpg)
Simple VoiceXML Example
![Page 35: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/35.jpg)
Frame-based Systems:Pros and Cons
Advantages
![Page 36: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/36.jpg)
Frame-based Systems:Pros and Cons
AdvantagesRelatively flexible input – multiple inputs, ordersWell-suited to complex information access (air)Supports different types of initiative
Disadvantages
![Page 37: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/37.jpg)
Frame-based Systems:Pros and Cons
AdvantagesRelatively flexible input – multiple inputs, ordersWell-suited to complex information access (air)Supports different types of initiative
Disadvantages Ill-suited to more complex problem-solving
Form-filling applications
![Page 38: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/38.jpg)
Dialogue Manager Tradeoffs
Flexibility vs Simplicity/PredictabilitySystem vs User vs Mixed Initiative
Order of dialogue interaction
Conversational “naturalness” vs Accuracy
Cost of model construction, generalization, learning, etc
![Page 39: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/39.jpg)
Dialog Systems DesignUser-centered design approach:
Study user and task:Interview users; record human-human interactions;
systems
![Page 40: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/40.jpg)
Dialog Systems DesignUser-centered design approach:
Study user and task:Interview users; record human-human interactions;
systems
Build simulations and prototypes:Wizard-of-Oz systems (WOZ): Human replaces system
Can assess issues in partial system; simulate errors, etc
![Page 41: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/41.jpg)
Dialog Systems DesignUser-centered design approach:
Study user and task:Interview users; record human-human interactions;
systems
Build simulations and prototypes:Wizard-of-Oz systems (WOZ): Human replaces system
Can assess issues in partial system; simulate errors, etc
Iteratively test on users: Redesign prompts (email subdialog)Identify need for barge-in
![Page 42: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/42.jpg)
SDS EvaluationGoal: Determine overall user satisfaction
Highlight systems problems; help tune
Classically: Conduct user surveys
![Page 43: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/43.jpg)
SDS EvaluationUser evaluation issues:
![Page 44: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/44.jpg)
SDS EvaluationUser evaluation issues:
Expensive; often unrealistic; hard to get real user to do
Create model correlated with human satisfaction
Criteria:
![Page 45: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/45.jpg)
SDS EvaluationUser evaluation issues:
Expensive; often unrealistic; hard to get real user to do
Create model correlated with human satisfaction
Criteria:Maximize task success
Measure task completion: % subgoals; Kappa of frame values
![Page 46: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/46.jpg)
SDS EvaluationUser evaluation issues:
Expensive; often unrealistic; hard to get real user to do
Create model correlated with human satisfaction
Criteria:Maximize task success
Measure task completion: % subgoals; Kappa of frame values
Minimize task costsEfficiency costs: time elapsed; # turns; # error correction
turnsQuality costs: # rejections; # barge-in; concept error rate
![Page 47: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/47.jpg)
PARADISE Model
![Page 48: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/48.jpg)
PARADISE ModelCompute user satisfaction with questionnaires
![Page 49: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/49.jpg)
PARADISE ModelCompute user satisfaction with questionnaires
Extract task success and costs measures from corresponding dialogsAutomatically or manually
![Page 50: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/50.jpg)
PARADISE ModelCompute user satisfaction with questionnaires
Extract task success and costs measures from corresponding dialogsAutomatically or manually
Perform multiple regression:Assign weights to all factors of contribution to UsatTask success, Concept accuracy key
![Page 51: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/51.jpg)
PARADISE ModelCompute user satisfaction with questionnaires
Extract task success and costs measures from corresponding dialogsAutomatically or manually
Perform multiple regression:Assign weights to all factors of contribution to UsatTask success, Concept accuracy key
Allows prediction of accuracy on new dialog w/Q&A
![Page 52: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/52.jpg)
Information State Dialogue Management
Problem: Not every task is equivalent to form-filling
Real tasks require:
![Page 53: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/53.jpg)
Information State Dialogue Management
Problem: Not every task is equivalent to form-filling
Real tasks require:Proposing ideas, refinement, rejection, grounding,
clarification, elaboration, etc
![Page 54: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/54.jpg)
Information State Dialogue Management
Problem: Not every task is equivalent to form-filling
Real tasks require:Proposing ideas, refinement, rejection, grounding,
clarification, elaboration, etc
Information state models include: Information state Dialogue act interpreterDialogue act generatorUpdate rulesControl structure
![Page 55: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/55.jpg)
Information State SystemsInformation state :
Discourse context, grounding state, intentions, plans.
![Page 56: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/56.jpg)
Information State SystemsInformation state :
Discourse context, grounding state, intentions, plans.
Dialogue acts:Extension of speech acts, to include grounding
actsRequest-inform; Confirmation
![Page 57: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/57.jpg)
Information State SystemsInformation state :
Discourse context, grounding state, intentions, plans.
Dialogue acts:Extension of speech acts, to include grounding
actsRequest-inform; Confirmation
Update rulesModify information state based on DAs
![Page 58: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/58.jpg)
Information State SystemsInformation state :
Discourse context, grounding state, intentions, plans.
Dialogue acts:Extension of speech acts, to include grounding
actsRequest-inform; Confirmation
Update rulesModify information state based on DAs
When a question is asked
![Page 59: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/59.jpg)
Information State SystemsInformation state :
Discourse context, grounding state, intentions, plans.
Dialogue acts:Extension of speech acts, to include grounding acts
Request-inform; Confirmation
Update rulesModify information state based on DAs
When a question is asked, answer itWhen an assertion is made,
![Page 60: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/60.jpg)
Information State SystemsInformation state :
Discourse context, grounding state, intentions, plans.
Dialogue acts:Extension of speech acts, to include grounding acts
Request-inform; Confirmation
Update rulesModify information state based on DAs
When a question is asked, answer itWhen an assertion is made,
Add information to context, grounding state
![Page 61: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/61.jpg)
Information State Architecture
Simple ideas, complex execution
![Page 62: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/62.jpg)
Dialogue ActsExtension of speech acts
Adds structure related to conversational phenomenaGrounding, adjacency pairs, etc
![Page 63: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/63.jpg)
Dialogue ActsExtension of speech acts
Adds structure related to conversational phenomenaGrounding, adjacency pairs, etc
Many proposed tagsetsVerbmobil: acts specific to meeting sched domain
![Page 64: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/64.jpg)
Dialogue ActsExtension of speech acts
Adds structure related to conversational phenomenaGrounding, adjacency pairs, etc
Many proposed tagsetsVerbmobil: acts specific to meeting sched domainDAMSL: Dialogue Act Markup in Several Layers
Forward looking functions: speech actsBackward looking function: grounding, answering
![Page 65: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/65.jpg)
Dialogue ActsExtension of speech acts
Adds structure related to conversational phenomenaGrounding, adjacency pairs, etc
Many proposed tagsetsVerbmobil: acts specific to meeting sched domainDAMSL: Dialogue Act Markup in Several Layers
Forward looking functions: speech actsBackward looking function: grounding, answering
Conversation acts:Add turn-taking and argumentation relations
![Page 66: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/66.jpg)
Verbmobil DA18 high level tags
![Page 67: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/67.jpg)
Dialogue Act Interpretation
Automatically tag utterances in dialogue
Some simple cases:Will breakfast be served on USAir 1557?
![Page 68: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/68.jpg)
Dialogue Act Interpretation
Automatically tag utterances in dialogue
Some simple cases:YES-NO-Q: Will breakfast be served on USAir
1557? I don’t care about lunch.
![Page 69: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/69.jpg)
Dialogue Act Interpretation
Automatically tag utterances in dialogue
Some simple cases:YES-NO-Q: Will breakfast be served on USAir
1557?Statement: I don’t care about lunch.Show be flights from L.A. to Orlando
![Page 70: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/70.jpg)
Dialogue Act Interpretation
Automatically tag utterances in dialogue
Some simple cases:YES-NO-Q: Will breakfast be served on USAir
1557?Statement: I don’t care about lunch.Command: Show be flights from L.A. to Orlando
Is it always that easy?Can you give me the flights from Atlanta to
Boston?
![Page 71: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/71.jpg)
Dialogue Act Interpretation
Automatically tag utterances in dialogue
Some simple cases:YES-NO-Q: Will breakfast be served on USAir 1557?Statement: I don’t care about lunch.Command: Show be flights from L.A. to Orlando
Is it always that easy?Can you give me the flights from Atlanta to Boston?
Syntactic form: question; Act: request/commandYeah.
![Page 72: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/72.jpg)
Dialogue Act Interpretation
Automatically tag utterances in dialogue
Some simple cases:YES-NO-Q: Will breakfast be served on USAir 1557?Statement: I don’t care about lunch.Command: Show be flights from L.A. to Orlando
Is it always that easy?Can you give me the flights from Atlanta to Boston?Yeah.
Depends on context: Y/N answer; agreement; back-channel
![Page 73: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/73.jpg)
Dialogue Act AmbiguityIndirect speech acts
![Page 74: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/74.jpg)
Dialogue Act AmbiguityIndirect speech acts
![Page 75: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/75.jpg)
Dialogue Act AmbiguityIndirect speech acts
![Page 76: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/76.jpg)
Dialogue Act AmbiguityIndirect speech acts
![Page 77: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/77.jpg)
Dialogue Act AmbiguityIndirect speech acts
![Page 78: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/78.jpg)
Dialogue Act RecognitionHow can we classify dialogue acts?
Sources of information:
![Page 79: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/79.jpg)
Dialogue Act RecognitionHow can we classify dialogue acts?
Sources of information:Word information:
Please, would you: request; are you: yes-no question
![Page 80: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/80.jpg)
Dialogue Act RecognitionHow can we classify dialogue acts?
Sources of information:Word information:
Please, would you: request; are you: yes-no questionN-gram grammars
Prosody:
![Page 81: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/81.jpg)
Dialogue Act RecognitionHow can we classify dialogue acts?
Sources of information:Word information:
Please, would you: request; are you: yes-no questionN-gram grammars
Prosody:Final rising pitch: question; final lowering: statementReduced intensity: Yeah: agreement vs backchannel
![Page 82: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/82.jpg)
Dialogue Act RecognitionHow can we classify dialogue acts?
Sources of information:Word information:
Please, would you: request; are you: yes-no questionN-gram grammars
Prosody:Final rising pitch: question; final lowering: statementReduced intensity: Yeah: agreement vs backchannel
Adjacency pairs:
![Page 83: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/83.jpg)
Dialogue Act RecognitionHow can we classify dialogue acts?
Sources of information:Word information:
Please, would you: request; are you: yes-no questionN-gram grammars
Prosody:Final rising pitch: question; final lowering: statementReduced intensity: Yeah: agreement vs backchannel
Adjacency pairs:Y/N question, agreement vs Y/N question, backchannelDA bi-grams
![Page 84: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/84.jpg)
Task & CorpusGoal:
Identify dialogue acts in conversational speech
![Page 85: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/85.jpg)
Task & CorpusGoal:
Identify dialogue acts in conversational speech
Spoken corpus: SwitchboardTelephone conversations between strangersNot task oriented; topics suggested1000s of conversations
recorded, transcribed, segmented
![Page 86: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/86.jpg)
Dialogue Act TagsetCover general conversational dialogue acts
No particular task/domain constraints
![Page 87: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/87.jpg)
Dialogue Act TagsetCover general conversational dialogue acts
No particular task/domain constraints
Original set: ~50 tags Augmented with flags for task, conv mgmt
220 tags in labeling: some rare
![Page 88: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/88.jpg)
Dialogue Act TagsetCover general conversational dialogue acts
No particular task/domain constraints
Original set: ~50 tags Augmented with flags for task, conv mgmt
220 tags in labeling: some rare
Final set: 42 tags, mutually exclusiveSWBD-DAMSLAgreement: K=0.80 (high)
![Page 89: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/89.jpg)
Dialogue Act TagsetCover general conversational dialogue acts
No particular task/domain constraints
Original set: ~50 tags Augmented with flags for task, conv mgmt
220 tags in labeling: some rare
Final set: 42 tags, mutually exclusiveSWBD-DAMSLAgreement: K=0.80 (high)
1,155 conv labeled: split into train/test
![Page 90: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/90.jpg)
Common Tags
Statement & Opinion: declarative +/- op
Question: Yes/No&Declarative: form, force
Backchannel: Continuers like uh-huh, yeah
Turn Exit/Adandon: break off, +/- pass
Answer : Yes/No, follow questions
Agreement: Accept/Reject/Maybe
![Page 91: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/91.jpg)
Probabilistic Dialogue Models
HMM dialogue models
![Page 92: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/92.jpg)
Probabilistic Dialogue Models
HMM dialogue modelsStates = Dialogue acts; Observations: Utterances
Assume decomposable by utteranceEvidence from true words, ASR words, prosody
![Page 93: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/93.jpg)
Probabilistic Dialogue Models
HMM dialogue modelsStates = Dialogue acts; Observations: Utterances
Assume decomposable by utteranceEvidence from true words, ASR words, prosody
![Page 94: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/94.jpg)
Probabilistic Dialogue Models
HMM dialogue modelsStates = Dialogue acts; Observations: Utterances
Assume decomposable by utteranceEvidence from true words, ASR words, prosody
![Page 95: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/95.jpg)
Probabilistic Dialogue Models
HMM dialogue modelsStates = Dialogue acts; Observations: Utterances
Assume decomposable by utteranceEvidence from true words, ASR words, prosody
![Page 96: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/96.jpg)
DA Classification - ProsodyFeatures:
Duration, pause, pitch, energy, rate, genderPitch accent, tone
Results:Decision trees: 5 common classes
45.4% - baseline=16.6%
![Page 97: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/97.jpg)
Prosodic Decision Tree
![Page 98: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/98.jpg)
DA Classification -WordsWords
Combines notion of discourse markers and collocations: e.g. uh-huh=Backchannel
Contrast: true words, ASR 1-best, ASR n-best
Results:Best: 71%- true words, 65% ASR 1-best
![Page 99: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/99.jpg)
DA Classification - AllCombine word and prosodic information
Consider case with ASR words and acoustics
![Page 100: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/100.jpg)
DA Classification - AllCombine word and prosodic information
Consider case with ASR words and acousticsProsody classified by decision trees
Incorporate decision tree posteriors in model for P(f|d)
![Page 101: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/101.jpg)
DA Classification - AllCombine word and prosodic information
Consider case with ASR words and acousticsProsody classified by decision trees
Incorporate decision tree posteriors in model for P(f|d)
Slightly better than raw ASR
![Page 102: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/102.jpg)
Integrated Classification
Focused analysisProsodically disambiguated classes
Statement/Question-Y/N and Agreement/BackchannelProsodic decision trees for agreement vs backchannel
Disambiguated by duration and loudness
![Page 103: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/103.jpg)
Integrated Classification
Focused analysisProsodically disambiguated classes
Statement/Question-Y/N and Agreement/BackchannelProsodic decision trees for agreement vs backchannel
Disambiguated by duration and loudness
Substantial improvement for prosody+wordsTrue words: S/Q: 85.9%-> 87.6; A/B: 81.0%->84.7
![Page 104: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/104.jpg)
Integrated Classification
Focused analysisProsodically disambiguated classes
Statement/Question-Y/N and Agreement/BackchannelProsodic decision trees for agreement vs backchannel
Disambiguated by duration and loudness
Substantial improvement for prosody+wordsTrue words: S/Q: 85.9%-> 87.6; A/B: 81.0%->84.7ASR words: S/Q: 75.4%->79.8; A/B: 78.2%->81.7
More useful when recognition is iffy
![Page 105: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/105.jpg)
Many VariantsMaptask: (13 classes)
Serafin & DiEugenio 2004Latent Semantic analysis on utterance vectorsText onlyGame information; No improvement for DA history
![Page 106: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/106.jpg)
Many VariantsMaptask: (13 classes)
Serafin & DiEugenio 2004Latent Semantic analysis on utterance vectorsText onlyGame information; No improvement for DA history
Surendran & Levow 2006SVMs on term n-grams, prosodyPosteriors incorporated in HMMs
Prosody, sequence modeling improves
![Page 107: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/107.jpg)
Many VariantsMaptask: (13 classes)
Serafin & DiEugenio 2004Latent Semantic analysis on utterance vectorsText onlyGame information; No improvement for DA history
Surendran & Levow 2006SVMs on term n-grams, prosodyPosteriors incorporated in HMMs
Prosody, sequence modeling improves
MRDA: Meeting tagging: 5 broad classes
![Page 108: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/108.jpg)
ObservationsDA classification can work on open domain
Exploits word model, DA context, prosodyBest results for prosody+wordsWords are quite effective alone – even ASR
Questions:
![Page 109: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/109.jpg)
ObservationsDA classification can work on open domain
Exploits word model, DA context, prosodyBest results for prosody+wordsWords are quite effective alone – even ASR
Questions: Whole utterance models? – more fine-grainedLonger structure, long term features
![Page 110: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/110.jpg)
Detecting Correction ActsMiscommunication is common in SDS
Utterances after errors misrecognized >2x as oftenFrequently repetition or paraphrase of original input
![Page 111: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/111.jpg)
Detecting Correction ActsMiscommunication is common in SDS
Utterances after errors misrecognized >2x as oftenFrequently repetition or paraphrase of original input
Systems need to detect, correct
![Page 112: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/112.jpg)
Detecting Correction ActsMiscommunication is common in SDS
Utterances after errors misrecognized >2x as oftenFrequently repetition or paraphrase of original input
Systems need to detect, correct
Corrections are spoken differently:Hyperarticulated (slower, clearer) -> lower ASR
conf.
![Page 113: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/113.jpg)
Detecting Correction ActsMiscommunication is common in SDS
Utterances after errors misrecognized >2x as oftenFrequently repetition or paraphrase of original input
Systems need to detect, correct
Corrections are spoken differently:Hyperarticulated (slower, clearer) -> lower ASR
conf.Some word cues: ‘No’,’ I meant’, swearing..
![Page 114: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/114.jpg)
Detecting Correction ActsMiscommunication is common in SDS
Utterances after errors misrecognized >2x as oftenFrequently repetition or paraphrase of original input
Systems need to detect, correct
Corrections are spoken differently:Hyperarticulated (slower, clearer) -> lower ASR
conf.Some word cues: ‘No’,’ I meant’, swearing..
Can train classifiers to recognize with good acc.
![Page 115: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/115.jpg)
Generating Dialogue ActsGeneration neglected relative to generation
![Page 116: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/116.jpg)
Generating Dialogue ActsGeneration neglected relative to generation
Stent (2002) model: Conversation acts, Belief modelDevelops update rules for content planning, e.g.
If user releases turn, system can do ‘TAKE-TURN’ actIf system needs to summarize, use ASSERT act
![Page 117: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/117.jpg)
Generating Dialogue ActsGeneration neglected relative to generation
Stent (2002) model: Conversation acts, Belief modelDevelops update rules for content planning, i.e.
If user releases turn, system can do ‘TAKE-TURN’ actIf system needs to summarize, use ASSERT act
Identifies turn-taking as key aspect of dialogue gen.
![Page 118: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/118.jpg)
Generating ConfirmationSimple systems use fixed confirmation strategy
Implicit or explicit
![Page 119: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/119.jpg)
Generating ConfirmationSimple systems use fixed confirmation strategy
Implicit or explicit
More complex systems can select dynamicallyUse information state and features to decide
![Page 120: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/120.jpg)
Generating ConfirmationSimple systems use fixed confirmation strategy
Implicit or explicit
More complex systems can select dynamicallyUse information state and features to decide
Likelihood of error: Low ASR confidence score
If very low, can reject
![Page 121: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/121.jpg)
Generating ConfirmationSimple systems use fixed confirmation strategy
Implicit or explicit
More complex systems can select dynamicallyUse information state and features to decide
Likelihood of error: Low ASR confidence score
If very low, can reject Sentence/prosodic features: longer, initial pause, pitch
range
![Page 122: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/122.jpg)
Generating ConfirmationSimple systems use fixed confirmation strategy
Implicit or explicit
More complex systems can select dynamicallyUse information state and features to decide
Likelihood of error: Low ASR confidence score
If very low, can reject Sentence/prosodic features: longer, initial pause, pitch
range
Cost of error:
![Page 123: Dialogue Management Ling575 Discourse and Dialogue May 18, 2011.](https://reader036.fdocuments.net/reader036/viewer/2022062407/56649d2e5503460f94a065b7/html5/thumbnails/123.jpg)
Generating ConfirmationSimple systems use fixed confirmation strategy
Implicit or explicit
More complex systems can select dynamicallyUse information state and features to decide
Likelihood of error: Low ASR confidence score
If very low, can reject Sentence/prosodic features: longer, initial pause, pitch range
Cost of error: Book a flight vs looking up information
Markov Decision Process models more detailed