Context-aware Neural-based Dialog Act Classification on Automatically\n Generated Transcriptions

2019/02/28 by Daniel Ortega, Chia-Yu Li, Ortega, Daniel +7
Computer Science · #Speech and dialogue systems #Multi-Agent Systems and Negotiation #Topic Modeling

paper · pdf · doi:10.48550/arxiv.1902.11060

Abstract

This paper presents our latest investigations on dialog act (DA)\nclassification on automatically generated transcriptions. We propose a novel\napproach that combines convolutional neural networks (CNNs) and conditional\nrandom fields (CRFs) for context modeling in DA classification. We explore the\nimpact of transcriptions generated from different automatic speech recognition\nsystems such as hybrid TDNN/HMM and End-to-End systems on the final\nperformance. Experimental results on two benchmark datasets (MRDA and SwDA)\nshow that the combination CNN and CRF improves consistently the accuracy.\nFurthermore, they show that although the word error rates are comparable,\nEnd-to-End ASR system seems to be more suitable for DA classification.\n

Related

• Neural-based Context Representation Learning for Dialog Act Classification
• Lexico-acoustic Neural-based Models for Dialog Act Classification
• Multi-level Gated Recurrent Neural Network for Dialog Act Classification
• End-to-end speech-to-dialog-act recognition
• Dialogue Act Classification with Context-Aware Self-Attention