[4Q2-GS-9] 自然言語処理・情報検索: 質問応答システム

[4Q2-GS-9-03] Semi-automatic Generation of Spoken English Dialogue Scenarios Based on Service Process Model

〇Emmanuel Ayedoun¹, Yuki Hayashi¹, Kazuhisa Seta¹ (1. Osaka Prefecture University)

キーワード：会話システム、サービス指向対話シナリオ、オーサリングシステム

The aim of this study is to ease the design of spoken English dialogue scenarios (i.e. the dialogue task structure and the dialogue flow logic), as a starting point towards providing English as a foreign language learners opportunities to simulate the target language in various computer-based realistic conversation situations. In this paper, we specifically focus on services (e.g. restaurant, hotel, travel booking, etc.) which seem to share a certain degree of structural similarity at the task level, and ultimately build an authoring tool that embeds a generic service task structure model in order to facilitate the implementation of services-oriented dialogues across various services domains. In such a way, dialogue scenario designers could just focus on specifying domain-specific aspects of the dialogue scenario, which is expected to lessen new scenarios implementation effort and speed up the design process of various service-oriented dialogue situations. We present a conceptual framework for dialogue scenario design which provides an overview of the proposed dialogue scenario authoring workflow.