Article Structure
Abstract
Substantial research effort has been invested in recent decades into the computational study and automatic processing of multiparty conversation.
Introduction
Substantial research effort has been invested in recent decades into the computational study and automatic processing of multiparty conversation.
Data
Analysis and experiments are performed using the ICSI Meeting Corpus (J anin et al., 2003; Shriberg et al., 2004).
Conceptual Framework
3.1 Definitions
Direct Estimation of 6)
Direct application of bi gram modeling techniques, defined over the states is treated as a baseline.
Limitations and Desiderata
As the analyses in Section 4 reveal, direct estimation can be useful under oracle conditions, namely when all of a conversation has been observed and the task is to find intervals where multi-participant behavior deviates significantly from its conversation-specific norm.
The Extended-Degree-of-Overlap Model
The limitations of direct models appear to be addressable by a form proposed by Laskowski and Schultz in (2006) and (2007).
Experiments
This section describes the performance of the discussed models on the entire ICSI Meeting Corpus.
Discussion
The model perpleXities as reported above may be somewhat different if the “talk spurt” were replaced by a more sociolinguistically motivated definition of “turn”, but the ranking of models and their relative performance differences are likely to remain quite similar.
Conclusions
This paper has presented a framework for quantifying the turn-taking perplexity in multiparty conversations.
Topics
language modeling
- The current work attempts to address this problem by proposing a simple framework, which, at least conceptually, borrows quite heavily from the standard language modeling paradigm.Page 1, “Introduction”
- In language modeling practice, one finds the likelihood P ( w | (9 of a word sequence w of length under a model (9, to be an inconvenient measure for comparison.Page 3, “Conceptual Framework”
- This makes it suitable for comparison of conversational genres, in much the same way as are general language models of words.Page 9, “Discussion”
- Accordingly, as for language models , density estimation in future turn-taking models may be im-Page 9, “Discussion”
See all papers in Proc. ACL 2010 that mention language modeling.
See all papers in Proc. ACL that mention language modeling.
Back to top.
bigram
- To produce Figures 1 and 2, a small fraction of probability mass was reserved for unseen bigram transitions (as opposed to backing off to unigram probabilities).Page 6, “Limitations and Desiderata”
- The EDO model mitigates R-specificity because it models each bigram (qt_1, qt) 2 (8,, S j) as the modified bigram (m, [0ij,nj]), involving three scalars each of which is a sum — a commutative (and therefore rotation-invariant) operation.Page 7, “The Extended-Degree-of-Overlap Model”
- Excluding qt_1 2 qt bigrams (leading to 0.32M frames from 2.39M frames in “all”) offers a glimpse of expected performance differences were duration modeling to be included in the models.Page 8, “Experiments”
See all papers in Proc. ACL 2010 that mention bigram.
See all papers in Proc. ACL that mention bigram.
Back to top.
