Presentation is loading. Please wait.

Presentation is loading. Please wait.

Dialogue Act Coding and Modalities GSLT: Dialogue Systems Leif Grönqvist – 11. June 2002 15:30.

Similar presentations


Presentation on theme: "Dialogue Act Coding and Modalities GSLT: Dialogue Systems Leif Grönqvist – 11. June 2002 15:30."— Presentation transcript:

1 Dialogue Act Coding and Modalities GSLT: Dialogue Systems Leif Grönqvist – leifg@ling.gu.se 11. June 2002 15:30

2 11 juni 2002Dialogue Systems: Leif Grönqvist2 Presentation Outline Properties for dialogue act (in particular) coding schemes Mode – medium – modality Modality Theory The different coding schemes Some interesting differences between the coding schemes Conclusions

3 11 juni 2002Dialogue Systems: Leif Grönqvist3 Properties for dialogue act coding schemes How general is it? Is it powerful enough for natural dialogue? Does the scheme handle different modalities? Are the definitions precise enough to make the scheme useful in dialogue systems?

4 11 juni 2002Dialogue Systems: Leif Grönqvist4 More properties for coding schemes Multi functional codings Mutual exclusive categories Discontinuous codings Relational codings Hierarchical coding values Multi-layer scheme

5 11 juni 2002Dialogue Systems: Leif Grönqvist5 Mode – medium – modality Some terms are used in different ways in different contexts Bretan and Bernsen use “modality” in the same way but psychologists do not. B & B do not agree on the term “medium” Bernsen: “We should aim for a terminology that is robust, conceptually clear and intuitively accepted.”

6 11 juni 2002Dialogue Systems: Leif Grönqvist6 Modality Theory Niels Ole Bernsen’s theory: “A generative taxonomy of output modalities” Start with a set of basic features: Linguistic/non-linguistic (non-)analogue (non-)arbitrary static/dynamic graphics/sound/touch

7 11 juni 2002Dialogue Systems: Leif Grönqvist7 Modality Theory 2 Combine them to get 48 distinct types Remove impossible combinations: 20 left One more feature: Real world/diagrammic/graphs resulting in 28 distinct modalities

8 11 juni 2002Dialogue Systems: Leif Grönqvist8

9 11 juni 2002Dialogue Systems: Leif Grönqvist9

10 11 juni 2002Dialogue Systems: Leif Grönqvist10 28 unimodal modalities Use of more than one will result in multimodality

11 11 juni 2002Dialogue Systems: Leif Grönqvist11 Selected Coding Schemes for Dialogue Acts LINLIN 1/2: Linköping, Ahrenberg et al, 1995 HCRC: Developed for the Map Task Corpus, Andersson et al 1991 DAMSL: By Discourse Resource Initiative as a standardized coding scheme, 1991 SWBD-DAMSL: Modified DAMSL by Stolcke et al 2000 GBG: Communicative Acts by Allwood 2000

12 11 juni 2002Dialogue Systems: Leif Grönqvist12 Why these They cover some different types And are developed for different purposes Some of them are widely spread and well known I know something about them

13 11 juni 2002Dialogue Systems: Leif Grönqvist13 Interesting differences LINLIN and DAMSL are more general than GBG and HCRC GBG and DAMSL are the more powerful DAMSL and HCRC do not handle non- verbal dialogue acts as well as LINLIN and GBG GBG is the only one not directly useful in dialogue systems

14 11 juni 2002Dialogue Systems: Leif Grönqvist14 Conclusions Some researchers does not seem to believe in non-verbal dialogue acts at all: in SWBD-DAMSL the coding types are mutually exclusive and two of the most common are: Backchannel/Acknowledge Non-verbal

15 11 juni 2002Dialogue Systems: Leif Grönqvist15 More conclusions The linguists scheme (GBG) is very rich but not useful I a dialogue system context Modality should not be used to define dialogue act categories – but in a second layer. Our intuition says that a nod or pointing at something could be an answer to a question

16 11 juni 2002Dialogue Systems: Leif Grönqvist16 We are done And probably out of time


Download ppt "Dialogue Act Coding and Modalities GSLT: Dialogue Systems Leif Grönqvist – 11. June 2002 15:30."

Similar presentations


Ads by Google