A two-stage LLM pipeline that first generates an explanation and then personalises it using a stored user knowledge memory reduces explanation length only when the user has related prior knowledge, in synthetic hospital and kitchen robot scenarios.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.RO 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Personalised Explanations in Long-term Human-Robot Interactions
A two-stage LLM pipeline that first generates an explanation and then personalises it using a stored user knowledge memory reduces explanation length only when the user has related prior knowledge, in synthetic hospital and kitchen robot scenarios.