Dialogue Management: Questions to Session 5
- Clarification:
I don't understand the following lines in (Traum:"Speech Acts ...", page
16, line 1): "Bunt notes, [...] that giving positive feedback at one
level implies problems at a higher level, while giving negative feedback
at one level implies success at lower levels."
- What are the pros and cons of the rank- vs. the level model of
actions? (section 4.2: Multiple Level of Interaction in Traum:"Speech
Acts for Dialogue Agents")
- In Question 5 to 8 (Traum's "20 Questions..."), Traum mentions the problems of guaranteeing that the speaker's
intention and the addressee's uptake agree. He quotes, for example the problem of
wrong speech recognition as a source of misinterpretation which may lead to additional
negotiations. Are there any other sources which lead to a different interpretation
of the common intention? If this is the case, which of them is the most problematic
in terms of dialogue management? I mean, the speech recognition problem should be
solved by an adequate speech recognizer and is not a problem due to poor dialogue
management.
- Witch are the advantadges or disadvantadges of the use of a lexical
semantic based / ontology of action based definition of dialog acts. Which
are the different applications of those issues?
- Why is a distinction of dialogue acts and others necessarry (ref Q11),
meaningful physical acts need multi agents, too.
- (How) Do we learn speech acts?
- How big are cross-cultural/linguistical differences?
/li>
- To what extent do animals use kinds of "speech acts"?
Ivana Kruijff-Korbayova
Last modified: Tue May 21 16:27:59 CEST 2002