This proposal relates to the technology of Augmentative and Alternative Communication (AAC). The research, to be developed over the three-year course of this project, relates to increasing communication speed for adult users of typing-based AAC devices. The proposed method has commonalities both with chatter bots and more sophisticated automated question answering systems. In particular, we propose to develop a program that will mine a very large database of stored interactions for sentences that are similar to the sentence currently being uttered by the interlocutor, and propose a set of plausible responses for the AAC user. The outcome of this research will be a system that improves over the current state of the art in whole utterance approaches in AAC, making use of sophisticated natural language processing techniques. Through this research and its practical application to helping real people with real communications needs, as well as coursework, seminars, participation in the AAC and disabilities community in Portland, OR, and intensive one-on-one meetings with his mentor Dr. Melanie Fried-Oken, the PI will accrue substantial clinical experience in AAC, and will gain a deep understanding of how technology can be used to help people.
The proposed project will develop a research program in the field of Augmentative and Alternative Communication. The program proposes to improve the throughput of AAC devices for conversation by modeling dialogue context for literate adult users. Specifically, we will use corpus-based techniques from question answering and chatter bots to select appropriate utterances in response to utterances from interlocutors.