トップ 差分 一覧 ソース 検索 ヘルプ RSS ログイン

Research

Development methodology of multi-modal dialogue system

In this research, we design and implement a new MMI description language named MrailsScript from the software engineering point of view. MrailsScript consists of the datamodel definition that can be inherited from semantic web class definition, and the annotations of interaction task and dialogue initiative. By expanding the existing web application development framework, our Mrails framework can automatically generate prototype codes of MMI application which is based on MVC (Model-View-Controller) model. We also implemented a helper application of MMI system development, MrailsBuilder, which assists coding of MrailsScript and realizes multilingual capability of the target MMI system.

  MrailsScript

Data modeling language for automatic prototyping of multi-modal dialogue system

 Mrails

Generator of the code for multi-modal dialogue system based on Grails web application framework

 MrailsBuilder

Assistant environment for coding MrailsScript using semantic web technology

 Publications

  • Daisuke Takegoshi and Masahiro Araki: Development Environment for Multimodal Interactive System based on Ontological Knowledge, In Proc. IIAI AAI 2014, pp.785-788, DOI 10.1109/IIAI-AAI.2014.158, 2014.
  • Masahiro Araki and Daisuke Takegoshi: A Rapid Development Framework for Multilingual Spoken Dialogue Systems, In Proc. COMPSAC 2012, pp.195-202, 2012.
  • Masahiro Araki: Rapid Development Process of Spoken Dialogue Systems using Collaboratively Constructed Semantic Resources, In Proc. SIGDial 2012, pp.70-73, 2012.
  • Masahiro Araki and Daisuke Takegoshi: Framework for the Development of Spoken Dialogue System based on Collaboratively Constructed Semantic Resources, In Proc. NAACL-HLT Workshop, pp.25-28, 2012.
  • Masahiro Araki and Yuko Mizukami: Development of a Data-driven Framework for Multimodal Interactive Systems, In Proceedings of IWSDS 2011, pp.91-101, 2011.

Architecture of Spoken dialogue system

Task-oriented dialogue agents efficiently solve a problem by communication. By incorporating open-domain conversation or chat functionality to the agent, it is possible for the user to increase their affection to the agent and to maintain the motivation of doing tasks. In addition, it may help the novice user to acquire the good usage of speech interface. However, it is generally difficult to combine a task-oriented dialogue system and a chat system because of the difference of dialogue management method of these two types. In addition, it is necessary to determine whether the user's utterance intends a chat dialogue or a task-oriented dialogue and which tasks are intended. Furthermore, if the user continues a chat dialogue in spite of the existence of an urgent task, the system needs to propose the task to the user. To solve this problem, we designed and implemented a development framework which has a switching functionality of a dialogue management with a little burden to developers. We examined the efficacy of the proposed framework via an example system development.