Formal specification description of multi-modal interface and its verification

多模态接口形式化规范描述及其验证

基本信息

批准号：
12308015
负责人：
INAGAKI Yasuyoshi
金额：
$ 20.25万
依托单位：
Nagoya University
依托单位国家：
日本
项目类别：
Grant-in-Aid for Scientific Research (A)
财政年份：
2000
资助国家：
日本
起止时间：
2000 至 2002
项目状态：
已结题

项目摘要

We advanced the research from the preceding year further, performed the formal description for the multi-modal interface in the current fiscal year, and developed the technique of its evaluation. Specifically, the following research topics were promoted.User model : we examined how to describe the user's behavior, toward the realization of a user model. Especially, we confirmed that describing correctly the behavior of a user, which becomes important in the principle of a user interface, allows more precise verification of the specification of a system. While collecting and analyzing the previous research result about the principle of a user interface, aiming at describing as behavior based on a user's knowledge, the subject experiment using the telephone etc. was conducted and these results were evaluated.Knowledge model : the knowledge model for expressing the knowledge state between a user and a system was examined. Since a knowledge model which can describe various task domains was … More asked, the task domain dialogues of the system developed so far were collected. The cooperation possibility with other models was confiremed by analyzing this model.Dialogue model : the dialogue model was analyzed using large-scale corpora, such as an in-car spoken dialogue database and a simultaneous interpretation database which are currently collected at the Center for Integrated Acoustic Information Research, Nagoya University. Statistical characteristic features were drawn and the formalization of the dialogue model in various situations was considered.Formal specification description : the multi-modal interface was described formally by making cooperate and complement of a user model, a knowledge model, and a dialogue model mutually. The technique of specification description obtained by the preceding year was summarized, and the verification work was prepared.Three kinds of above-mentioned models were analyzed in detail, a still more formal description was defined, and the foundation of the evaluation method was established. The possibility of the comparison of a system and the evaluation, and the reuse to an advancement was confirmed by formalizing various interfaces by universal specification description. Less

我们进一步推进了上一年的研究，对本财年的多模态界面进行了形式化描述，并开发了其评估技术。具体来说，推进了以下研究主题。用户模型：我们研究了如何。描述用户的行为，以实现用户模型，特别是，我们确认正确描述用户的行为（这在用户界面的原理中变得很重要）可以在收集时更精确地验证系统的规范。并对前人关于用户原理的研究成果进行分析界面，旨在描述基于用户知识的行为，进行了使用电话等的主题实验，并对这些结果进行了评估。知识模型：用于表达用户和系统之间的知识状态的知识模型。提出了一个可以描述各种任务领域的知识模型，通过分析该模型，收集了迄今为止开发的系统的任务域对话，确认了与其他模型的合作可能性。对话模型：使用大型对话模型进行了分析。 -规模语料库，例如绘制了名古屋大学综合声学信息研究中心目前收集的车载语音对话数据库和同声传译数据库的统计特征，并考虑了各种情况下对话模型的形式化。形式化规范描述：通过用户模型、知识模型和对话模型的相互配合和补充，对多模态界面进行了形式化描述，总结了前一年获得的规范描述技术，并做好了验证工作。三种类型。对上述模型进行了分析具体来说，定义了更正式的描述，并通过通用规范描述将各种接口形式化，确定了系统比较和评估以及重用到进步的可能性。较少的