DMCA
The Effects of Disfluency Detection in Parsing Spoken Language
Cached
Download Links
| Citations: | 1 - 0 self |
BibTeX
@MISC{Jørgensen_theeffects,
author = {Fredrik Jørgensen},
title = {The Effects of Disfluency Detection in Parsing Spoken Language},
year = {}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
Spoken language contains disfluencies that, because of their irregular nature, may lead to reduced performance of data-driven parsers. This paper describes an experiment that quantifies the effects of disfluency detection and disfluency removal on data-driven parsing of spoken language data. The experiment consists of creating two reduced versions from a spoken language treebank, the Switchboard Corpus, mimicking a speechrecognizer output with and without disfluency detection and deletion. Two datadriven parsers are applied on the new data, and the parsers ’ output is evaluated and compared. 1
Keyphrases
disfluency detection parsing spoken language speechrecognizer output parser output switchboard corpus data-driven parser spoken language treebank reduced performance disfluency removal spoken language data reduced version data-driven parsing datadriven parser new data experiment consists spoken language irregular nature
