PEC-Home: Interpretation of Progressively Elliptical Commands in Smart Homes

PEC-Home is a newly introduced simulated dataset aimed at improving the interpretation of progressively elliptical commands in smart home environments. It addresses two main challenges: referential ambiguity among multiple users and intention ambiguity due to changing user preferences. Experiments with various LLMs, including GPT-4o, reveal that current home assistants struggle with execution accuracy when relying solely on elliptical commands, highlighting the need for enhanced contextual understanding in AI-driven home automation systems.