Abstract

Assisting users with To Do lists presents new challenges for intelligent user interfaces. This paper presents our approach and an implemented system, BEAM, to process To Do list entries and map them to tasks that can be automated for the user. The system then monitors the progress of the execution of those tasks and processes their completion. We use a novel approach to interpreting natural language tasks that exploits semiformal knowledge repositories collected from web volunteers, which are broad-coverage and continuously increase in size at zero cost. An important aspect of our research is that it has been heavily influenced by an office assistant architecture that learns continuously to assist the user with new tasks.

Date

March 7, 2026

Authors

Yolanda Gil, Timothy Chklovski

Conference

AAAI Spring Symposium: Interaction Challenges for Intelligent Assistants

Pages

43-51