Articulate+ : An Always-Listening Natural Language Interface for Creating Data Visualizations
Articulate+ : An Always-Listening Natural Language Interface for Creating Data Visualizations
复制标题
DOI：
10.1145/3543829.3544534
复制
发表时间：
2022-07
期刊：
影响因子：
--
通讯作者：
Roderick S. Tabalba;Nurit Kirshenbaum;J. Leigh;Abari Bhattacharya;Andrew E. Johnson;Veronica Grosso;Barbara Maria Di Eugenio;Moira L. Zellner
中科院分区：
文献类型：
--
作者：
Roderick S. Tabalba;Nurit Kirshenbaum;J. Leigh;Abari Bhattacharya;Andrew E. Johnson;Veronica Grosso;Barbara Maria Di Eugenio;Moira L. Zellner
用于表达数据可视化的自然语言接口和语音用户接口面临着歧义，如语音不流畅、不规范和缩写。在本文中，我们描述了一个始终在倾听的人工智能代理，它可以消除请求的歧义，同时也可以自发地呈现信息可视化。我们进行了初步的用户研究，以深入了解为数据可视化提供始终监听的界面所涉及的问题。我们的早期结果表明，通过利用伶俐+的始终倾听的界面，用户能够通过更少的查询获得他们想要的可视化，同时也能够从系统生成的自发可视化中受益。
Natural Language Interfaces and Voice User Interfaces for expressing data visualizations face ambiguities, such as, speech disfluency, under-specification, and abbreviations. In this paper, we describe Articulate+, an Artificial Intelligence Agent that is always listening, built to disambiguate requests while also spontaneously presenting informative visualizations. We conducted a preliminary user study to gain insight into the issues involved in providing an always-listening interface for data visualization. Our early results suggest that by leveraging Articulate+’s always-listening interface, users are able to obtain their desired visualizations with fewer queries while also being able to benefit from spontaneous visualizations generated by the system.