MultiTurnCleanup: A Benchmark for Multi-Turn Spoken Conversational Transcript Cleanup
Abstract
Current disfluency detection models focus on individual utterances each from a single speaker. However, numerous discontinuity phenomena in spoken conversational transcripts occur across multiple turns, hampering human readability and the performance of downstream NLP tasks. This study addresses these phenomena by proposing an innovative Multi-Turn Cleanup task for spoken conversational transcripts and collecting a new dataset, MultiTurnCleanup1. We design a data labeling schema to collect the high-quality dataset and provide extensive data analysis. Furthermore, we leverage two modeling approaches for experimental evaluation as benchmarks for future research.
Cite
@article{arxiv.2305.12029,
title = {MultiTurnCleanup: A Benchmark for Multi-Turn Spoken Conversational Transcript Cleanup},
author = {Hua Shen and Vicky Zayats and Johann C. Rocholl and Daniel D. Walker and Dirk Padfield},
journal= {arXiv preprint arXiv:2305.12029},
year = {2023}
}
Comments
EMNLP 2023 main conference. Dataset: https://github.com/huashen218/MultiTurnCleanup