Dissertation Defense - Mingrui Zhang
Towards Humanlike Communication with Computers
Communication is fundamental to human experience. It is also fundamental to our interaction with computers. We both communicate with computers and use computers to communicate with other people. Whether we’re on a desktop or mobile device, typing on keyboards remains the most common way of expressing our thoughts. However, there are often times our keyboards make mistakes and barely understand us. Therefore, an important question is how to make the keyboard understand our languages like a human being, so that communication with computers can be more efficient and natural?
In this talk, I will demonstrate how to design, build and evaluate humanlike communication interactions using the power of machine learning and natural language processing. I will specifically demonstrate two promising directions: (1) Intelligent text entry and editing interactions that utilize language context to understand the user’s intentions; and, (2) Assistive systems that help blind or low vision (BLV) users to communicate with emojis and animated GIFs online; (3) Models and metrics for evaluating intelligent information input systems. Together, the work demonstrates the statement: Artificial intelligence can enable and improve advanced text production and accessible interactions with pictures; in addition, new metrics for text entry can enable the evaluation of advanced capabilities.
Supervisory Committee:
Chair: Jacob O. Wobbrock, Professor, Information School
GSR: James Fogarty, Professor, Computer Science & Engineering
Member: Alexis Hiniker, Assistant Professor, Information School
Member: Leah Findlater, Associate Professor, Human Centered Design & Engineering
Member: Shumin Zhai, Principle Scientist, Google