Work

Task-Guiding Voice Assistants: Targeting Uncertainty

Year: 2022
Type: dissertation
Abstract: From the rise in popularity of conversational voice assistants such as Siri, Cortana and Alexa came the new challenge of assisting users with complex, multi-step, and multimodal tasks. Hence, the recent interest in task-guiding voice assistants. Task-guiding voice assistants aim to aid users to accomplish complex tasks by dissecting said tasks into comprehensive and accomplishable steps. However, they must still keep the conversational nature desired by a voice assistant. The development of task-guiding conversational voice assistants is required to tackle a wide variety of problems. These range from detecting what the user wishes to accomplish from the whole voice input to pre-processing the available documents and adapting the search results with given restrictions. Nevertheless, a very crucial aspect that might be easily overlooked is the software development aspects related to the project. This overlook may be the downfall of voice-assistant applications, or at least, make them very difficult to evolve. Furthermore, these systems are faced with a troublesome challenge: uncertainty. In this context, uncertainty limits development in two dimensions. The first is due to the unknown interaction flow caused by the conversational nature of systems in the domain. The second dimension is due to the domain is still being researched, making it difficult to determine what should be handled by the scope of the domain. The main objective of this dissertation is to propose a systematic software development approach, focusing on its early stages (i.e., requirements elicitation, analysis and specification, and architectural design), in the “Task Guiding Conversational Voice Assistants” domain. This dissertation also evaluates the contributions made to systems in the domain. These evaluations assure the benefit provided by the elaborated work. The evaluations focus on analyzing the impact the developed domain model, requirement specification approach, and architecture have on the domain at hand.
Language: English
Open Access status: green
OpenAlex ID: W7062408441
Cites:
Cited by:
Related to: 10