🤖 AI Summary
This work addresses the limitations of conversational AI systems in maintaining persistent memory across sessions, which are often constrained by context window sizes or susceptible to privacy leaks. To overcome these challenges, the authors propose a localized graph-based memory system that integrates a bitemporal model—capturing both valid and transaction time—with a semantic retrieval mechanism. By leveraging immutable memory nodes and automatically constructed semantic edges, the system enables efficient memory association and historical state tracing. Implemented using Neo4j property graphs, HNSW vector indexing, and 1024-dimensional embeddings, it supports point-in-time semantic queries without overwriting past records. Evaluation on the LongMemEval benchmark demonstrates strong performance, achieving 46.7% R@10 for current state retrieval and 80% accuracy on knowledge-update questions, thereby validating the efficacy of the proposed approach.
📝 Abstract
Conversational AI agents commonly lack persistent memory across sessions. The obvious fixes like injecting full chat histories into the context window, or delegating to a third-party memory service, either exhaust the model's context budget or send personal data through infrastructure the user does not control. We describe a memory store that avoids both problems: an agent-local Neo4j property graph augmented with HNSW vector indexes and a full bitemporal data model. Each memory is stored as an immutable identity node linked to versioned content nodes carrying two closed-open time intervals: valid time (when the fact was true in the world) and transaction time (when the database recorded it). This design supports point-in-time semantic retrieval without physically overwriting history. Semantic edges between related memories are maintained automatically at write time using cosine similarity over 1024-dimensional embeddings. We evaluate the system on LongMemEval, a 500-question benchmark spanning six question types designed to stress long-term memory. Across 60 sampled questions, the current-state semantic search path achieves 46.7% R@10 overall, rising to 80% on knowledge-update questions. The time-travel path yields 80% R@10 on knowledge-update but decreases recall on temporal-reasoning questions (50% to 37.5%), a consequence of post-filter dilution that points directly to a concrete design improvement. We discuss what these results reveal about the limits of pure retrieval for different question types and what each failure mode suggests for future work.