Dialog state tracking for unseen values using an extended attention mechanism

Takami Yoshida, Kenji Iwata, Hiroshi Fujimura, Masami Akamine

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Recently, discriminative models using recurrent neural networks (RNNs) have shown good performance for dialog state tracking (DST). However, the models have difficulty in handling new dialog states unseen in model training. This paper proposes a fully data-driven approach to DST that can deal with unseen dialog states. The approach is based on an RNN with an attention mechanism. The model integrates two variants of RNNs: a decoder that detects an unseen value from a user’s utterance using cosine similarity between word vectors of the user’s utterance and that of the unseen value; and a sentinel mixture architecture that merges estimated dialog states of the previous turn and the current turn. We evaluated the proposed method using the second and the third dialog state tracking challenge (DSTC 2 and DSTC 3) datasets. Experimental results show that the proposed method achieved DST accuracy of 80.0% for all datasets and 61.2% for only unseen dataset without hand-crafted rules and re-training. For the unseen dataset, the use of the cosine similarity-based decoder leads to a 26.0-point improvement from conventional neural network-based DST. Moreover, the integration of the cosine similarity-based decoder and the sentinel mixture architecture leads to a further 2.1-point improvement.

Original languageEnglish
Title of host publication9th International Workshop on Spoken Dialogue System Technology, IWSDS 2018
EditorsLuis Fernando D’Haro, Rafael E. Banchs, Haizhou Li
PublisherSpringer
Pages77-89
Number of pages13
ISBN (Print)9789811394423
DOIs
Publication statusPublished - 2019 Jan 1
Externally publishedYes
Event9th International Workshop on Spoken Dialogue System Technology, IWSDS 2018 - Singapore, Singapore
Duration: 2018 Apr 182018 Apr 20

Publication series

NameLecture Notes in Electrical Engineering
Volume579
ISSN (Print)1876-1100
ISSN (Electronic)1876-1119

Conference

Conference9th International Workshop on Spoken Dialogue System Technology, IWSDS 2018
CountrySingapore
CitySingapore
Period18/4/1818/4/20

ASJC Scopus subject areas

  • Industrial and Manufacturing Engineering

Fingerprint Dive into the research topics of 'Dialog state tracking for unseen values using an extended attention mechanism'. Together they form a unique fingerprint.

Cite this