Dialogue act annotation schemes are used to mark important characteristics of utterances. These annotations indicate the role of an utterance in a specific dialogue and make the relationship between utterances more obvious.
Most of the dialogue schemes nowadays are task-oriented as we will see later on in this report. This is due to reduce the amount of annotation tags to a capable size for annotation and to increase the analyzing rate of the NLP system in which the scheme is used. The information content (or the semantics) of task-oriented dialogues can be basically splitted into task / domain related information and information that addresses the communication process. To guarantee generality and therefore more flexibility both information levels should be kept separately in the notion-choice of tags. Schemes which cover only those two fields are said to be schemes for rather shallow analysis.
As an example of a scheme that allows deep analysis DAMSL can be mentioned.
With its forward and backward looking functions it keeps track of how an
utterance constrains the future beliefs and actions of the participants,
and affects the discourse and how an utterance relates to the previous
discourse, respectively.
Another reason might be that they are not used very much which could, for example, lead to the assumption that they are too complicated.
However, in the following some criteria for schemes are detailed which can be used to scale the observed schemes.
TR: travel
TS: transport
COS: computer operating systems
COU: courtroom interaction
BA: business appointments
DES: directory enquiry services
FUR: furnishing rooms interactively
DIR: giving directions
INST: giving instructions (e.g. about cooking)
CN: cooperative negotiation
IE: information extraction
PS: problem solving
TI: teaching/instruction
CO: counselling
CH: chatting
|
|
|
|
|
|
|
yes | yes | yes | yes | |
Annotators | Number | 3 | huge | 10 | 2 |
Expertise | experts | experts | experts | experts | |
Information about annotated dialogues | Size | 500 dialogues | 160MB | 22 dialogues | 16 dialogues |
Languages | Dutch | many | Japanese | English | |
Participants | 2 | 2 | 2 | 2 | |
Task Orientation | TD | (NTD) | TD | TD | |
Application Orientation | yes | no | no | yes | |
Domain Restriction | DES | no | DIR, BA, TR | FUR | |
Activity Type | IE | CH | CN, PS | CN | |
Human / Machine Part. | HH, MM | HH, NMM | HH, NMM(?) | HH, MM (computer) | |
|
yes (77% agreement) | no | yes (0.57 < alpha < 0.68) | yes | |
|
yes, own | yes, own | yes, SGML-like | yes, Nb | |
|
yes, OVR coder | yes | yes, modification of dat | yes, Nb | |
|
yes | no | ? | yes |
|
CECH |
|
|
|
|
|
yes | yes | yes | yes | |
Annotators | Number | 5 | 5 | 4 | 7 |
Expertise | fairly experts | experts | experts | trained | |
Information about annotated dialogues | Size | 88 dialogues | 230 dialogues | 18 dialogues | 25 dialogues |
Languages | English | Engl.,
Jap.,
Kor., It. |
English | English | |
Participants | 2 | 2 | 2 | 2 | |
Task Orientation | TD | TD | NTD | TD | |
Application Orientation | yes | yes | no | yes | |
Domain Restriction | TS | TR | no | DES | |
Activity Type | CN | CN | several | IE | |
Human / Machine Part. | HH, MM, NMM | HH | HH | HH, MM | |
|
yes (91% agreement) | no | yes, k =0.56 | yes, k =0,6+ | |
|
yes, Nb?s | yes | yes, DAMSL | yes | |
|
yes, Nb | no | yes, dat | yes | |
|
yes | yes | yes | ? |
|
|
|
|
|
|
|
yes | yes | yes | yes | |
Annotators | Number | 4 | 4 | 4 | 6 |
Expertise | experts | experts | experts | naive | |
Information about annotated dialogues | Size | many | 140 dialogues | 128 dialogues | 72 dialogues |
Languages | English | Swedish | English | English | |
Participants | 2 | 2 | 2 | 1 | |
Task Orientation | TD | TD | TD | TD | |
Application Orientation | yes | yes | yes | no | |
Domain Restriction | BA | TR/TS | DIR | INSTR | |
Activity Type | CN | IE | PS | TI | |
Humpan
/ Machine
Part. |
HH | HM, NS | HH, NMM | HH, NMM | |
|
yes (89% agreement) | yes (97% agreement) | Yes, k =0.83 | no | |
|
yes, own | yes, Nb?s | Yes, own SGML based | yes, Nb?s | |
|
no | yes, Nb | Yes, own | yes, Nb | |
|
yes | yes | yes | yes |
|
|
|
|
|
|
|
yes | yes | yes | yes | |
Annotators | Number | 7 | 9 | 3 | 3 |
Expertise | experts | experts | experts | naive | |
Information about annotated dialogues | Size | 100 dialogues | 1155 dialogues | 36 dialogues | 1172 dialogues |
Languages | Swedish | English | English | Eng., Jap., Ger. | |
Participants | 2 (?) | 2 | 2 | 2 | |
Task Orientation | TD | NTD | NTD | TD | |
Application Orientation | yes | no | yes | yes | |
Domain Restriction | COU | no | no | BA | |
Activity Type | several | several | CN | CN | |
Human
/
Machine Part. |
HH, NMM | HH, MM | HH, NM | HH, NMM | |
|
yes (not published) | yes, 0.8 < k < 0.84 | yes (not published) | yes, k =0.84 | |
|
yes, own | yes, variant of DAMSL | yes, Nb?s | yes, own | |
|
yes, TRACTOR | no | yes, Nb | yes, AnnoTag | |
|
yes | yes | yes | yes |
In order to develop a standard it is necessary
to compare schemes with regard to their underlying task and dialogue acts.
The following tables present domain-grouped schemes and show the equivalence
between their dialogue acts. Entries with italic font represent higher
order expressions which can't be annotated.
Domain: information retrieval
|
|
|
Moves
|
Speech Acts | Initiative |
Statement |
|
Update |
Question
Check Alignment |
Question-Confirm | Question |
|
Response | Response |
Clarification |
|
Answer |
|
Confirm
Accept Reject |
|
Acknowledgement
Reconfirmation |
Acknowledge
Repeat |
|
Greeting
Bye |
|
Discourse
Management
Opening Ending Continuation |
Pause |
|
|
Other |
|
|
Domain: route direction
|
|
Initiation | Initiating moves |
Inform
Other assertion |
Explain |
Yes-no
question
Wh-question |
Query-yn
Query-w Check Align |
Request
Suggest Persuasion Propose Demand |
Instruct |
Promise |
|
Response | Response moves |
Positive
Negative Answer Other response |
Reply-y
Reply-n Reply-w Clarify |
Hold
Confirm |
|
|
Acknowledge |
Follow-up
Understanding |
|
Conventional
Opening Closing |
|
Other initiation |
|
Domain: appointment scheduling
|
|
Initiation | Dialogue_Act |
Inform
Other assertion |
Inform
Init Give-reason Digress Deviate_Szenario Refer_to_setting |
Yes-no-question
Wh-question |
|
Request
Suggest Persuasion Propose Demand |
Suggest
Request Request_Suggest Request_Clarify Request_Comment |
Promise |
|
Response |
|
Positive
Negative Answer Other response |
Feedback
Feedback_Positive Feedback_Negative Feedback_Backchanneling |
Hold
Confirm |
Accept
Confirm Reject Explained_Reject Clarify Clarify_Answer |
Follow-up
Understanding |
|
Conventional
Opening Closing |
Convention
Thank Deliberate Introduce Politeness_Formula Greeting Greeting_Begin Greeting_End |
Other initiation | Not_Classifiable |
Domain: general
|
|
|
|
Forward looking function | Forward Communicative -Function | Illocutionary Function | Categories of Illocutionary Force |
Statement
Assert Reassert Other |
Statement
Statement-no-opinion Statement-opinion |
Inform
Supp-Inf Supp-Sug |
Statement:
AC, CN, DW, ST, WS Declarations: DC, DP |
Info-Request | Influencing-Addressee-Future-Action
(1)
Yes-No-Question Wh-Question Or-Clause Declarative-Yes-No-Question Declarative-Wh-Question Tag-Question Backchannel-in-Question Rhetorical-Question |
YNO
WHQ |
Questions:
AQ, AA, AN, EQ, NA, QA, QN, RA, SA, TA, TQ, YQ, RQ |
Influencing-Addressee-Future-Action
Action-directive Open-Option |
Influencing-Addressee-Future-Action
(2)
Open-Question Action-Directive |
Request
Suggest |
Directives
(1):
RP, RQ |
Committing-Speaker-Future-Action
Offer Commit Explicit-performative Exclamation |
Committing-Speaker-Future-Action
Offers Options Commits |
Offer | Commitments:
FP, PF, SI, TD Directives (2): CL, SS |
- | - | Promise | PD |
Backward looking function | Backwards-Communicative-Function | - | - |
Answer | Answer
Yes Answer No Answer Affirmative non-yes-answer Negative non-no answer Other answer Dispreferred answers |
Eval | Evaluations:
AB, CR, DS, ED, ET, PM Directives (3): AC |
Agreement
Accept Accept-part Maybe Reject Reject-part Hold |
Agreement
Agree/Accept Maybe / Accept-part Reject Hold before answer/agreement |
Accept
Reject Check |
Directives
(4):
AD, AL, CS, RD, GI, GR, DR Declarations (2): ND, YD |
Understanding | Understanding | Grounding | - |
- | - | RequestAck | - |
Signal-understanding
Acknowledge Repeat-rephrase Completion |
Response-Acknowledgement
Repeat-phrase Collaborative-Completion Acknowledge Summarize/Re-formulate Appreciation Downplayer |
Acknowledge | Speech
Elicitations:
CX, EA, EI, EC, EX, RT, SC |
Signal-Non-Understanding | Signal-non-understanding | Request-Repair | Demands
for clarification:
RR |
Correct-Misspeaking | Repair | Text
editing:
CT |
|
- | Other-forward-function
Conventional-opening Conventional-closing Thanking Apology |
Greet
Apologise |
- |
- | - | - | - |
- | Other
Quotation Hedge |
- | Vocalisation:
YY, OO |
- | - | - | Markings:
CM, EM, EN, ES, MK, TO, XA |
- | - | - | Performances:
PR, TX |
The huge amount of coding schemes detailed in the Annexes shows the current research interests in dialogue act annotation schemes. There also seems to be a trend to shallow, task-oriented annotation as these schemes predominate those which focus on a general approach. The comparison of dialogue acts of schemes with equivalent domains reflect the similarities expected. But more surprisingly even a dialogue act comparison among all schemes regardless to their orientation shows quite a lot of parallelism although the general schemes are, of course, more comprehensive.
In order to decide, which schemes should be taken under consideration in MATE, schemes should have a coding book, they should be heavily used and should have good evaluation results. Also a scheme that is not related to a special task seems to be more appropriate as a task related and therefor possibly restricted one. If we look at the general comparison of schemes above, one can observe that all listed schemes provide a coding book. Amongst the schemes which are mostly used we can see Alparon, Chat, SWBD-DAMSL and Verbmobil. Unfortunately Chat hasn't been evaluated, but Alparon, SWBD-DAMSL and Verbmobil are judged to be good. As SWBD-DAMSL is the only one of these schemes which is not task related, it should be definitely supported in MATE. With regard to the MATE standard, of course, the dialogue acts of all schemes should be taken into account and analysed.
Last Modification: 26.8.1998 by Marion
Klein