TU Darmstadt / ULB / TUbiblio

Ontology-based contextual coherence scoring

Porzel, Robert and Gurevych, Iryna and Müller, Christof (2003):
Ontology-based contextual coherence scoring.
In: Proceedings of the Fourth SIGdial Workshop on Discourse and Dialogue, [Conference or Workshop Item]

Abstract

In this paper we present a contextual extension to ONTOSCORE, a system for scoring sets of concepts on the basis of an ontology. We apply the contextually enhanced system to the task of scoring alternative speech recognition hypotheses (SRH) in terms of their semantic coherence. We conducted several annotation experiments and showed that human annotators can reliably differentiate between semantically coherent and incoherent speech recognition hypotheses (both with and without discourse context). We also showed, that annotators can reliably identify the overall best hypothesis from a given n-best list. While the original ONTOSCORE system correctly assigns the highest score to 84.06% of the corpus, the inclusion of the conceptual context increases the number of correct classifications to yield 86.76%, given a baseline of 63.91% in both cases.

Item Type: Conference or Workshop Item
Erschienen: 2003
Creators: Porzel, Robert and Gurevych, Iryna and Müller, Christof
Title: Ontology-based contextual coherence scoring
Language: German
Abstract:

In this paper we present a contextual extension to ONTOSCORE, a system for scoring sets of concepts on the basis of an ontology. We apply the contextually enhanced system to the task of scoring alternative speech recognition hypotheses (SRH) in terms of their semantic coherence. We conducted several annotation experiments and showed that human annotators can reliably differentiate between semantically coherent and incoherent speech recognition hypotheses (both with and without discourse context). We also showed, that annotators can reliably identify the overall best hypothesis from a given n-best list. While the original ONTOSCORE system correctly assigns the highest score to 84.06% of the corpus, the inclusion of the conceptual context increases the number of correct classifications to yield 86.76%, given a baseline of 63.91% in both cases.

Title of Book: Proceedings of the Fourth SIGdial Workshop on Discourse and Dialogue
Divisions: 20 Department of Computer Science
20 Department of Computer Science > Telecooperation
Date Deposited: 31 Dec 2016 12:59
Identification Number: porzel2003
Export:
Suche nach Titel in: TUfind oder in Google

Optionen (nur für Redakteure)

View Item View Item