DrJonathan Kummerfeld

I work on making computers capable of understanding language and using that ability to collaborate with people more effectively. This involves solving challenges in Artificial Intelligence and Human-Computer Interaction.

My research projects fall into three core areas:

Interactive Executable Semantic Parsing - Developing systems that can take complex actions in response to user requests. For example, a person asks for information and the system queries a database to provide an answer, which involves accurately interpreting the request and determining the best way to present the answer.

Data Science - Making it possible to use NLP technology to analyse new forms of data, broadening the range of questions we can answer. Real-world data is messier than the benchmarks used in most NLP research, posing new challenges for our methods.

Communication and Collaboration - Exploring how AI systems can be active collaborators with people. For example, by building bots to play the board game Diplomacy, in which players chat to negotiate alliances, strategies, and tactics.

Solving these challenges involves innovation in (1) AI models, including neural networks, (2) data creation methods, such as crowdsourcing, and (3) research methodology, e.g., metrics for performance.