With the current popularity of Smart Speaker like Amazon Alexa or Google Home, more and more people are familiar with the technology and are using them on the daily basis. However, most of the tasks that are performed on those devices are simple ones that only required one or two commands. According to Ruan (2016), interacting with a Voice User Interface (VUI) is three times faster and produces fewer errors than with a Graphic User Interface (GUI) when it comes to texting. As a result, voice devices have a huge potential for assisting humans in the future with their complicated tasks. One of the key factors for the system to perform more such jobs is how well the system can handle the error made by either the user or the system. By leveraging the technology of Amazon Alexa Developer Kits, I hope to conduct an experiment that can resemble a real-world situation where users interact with Conversation Agents. A skill will be developed on Amazon Alexa platform that allows users to set up a meeting with a person of their choice. The skill will then find the best mutual time between two people calendar to give a suggestion for the user. The experiment will be conducted at Union College using Amazon Alexa device. The data will be collected and investigated to find the optimal one among multiple error recovery strategies.
Reference:
Ruan, Sherry, et al. "Speech is 3x faster than typing for English and mandarin text entry on mobile devices." arXiv preprint arXiv:1608.07323 (2016).