AdaMem: Adaptive User-Centric Memory for Long-Horizon Dialogue Agents
Overview
We propose AdaMem, an adaptive user-centric memory framework for long-horizon dialogue agents. AdaMem organizes dialogue history into working, episodic, persona, and graph memories, and employs a question-conditioned retrieval route combining semantic retrieval with relation-aware graph expansion. AdaMem achieves state-of-the-art performance on the LoCoMo and PERSONAMEM benchmarks.