RuleMem: Active Rule Memory for Long-Term Conversational Agents
Xingyuan Zeng and colleagues propose RuleMem, which induces reusable natural-language Horn clauses from conversation history so that agent memory actively guides retrieval and reasoning instead of sitting as passively stored facts.
Ask this paper
Rules, not facts: memory entries are logical rules validated by a Rule Perplexity Consistency check, which lets the system retrieve semantically distant evidence that embedding similarity would never surface.
Closes the semantic gap: passive fact stores fail when the question and the evidence share no surface overlap. An induced rule supplies the bridging structure.
Large margin on LoCoMo: highest accuracy against 14 baselines, exceeding the baseline average by 27.47 points, a 54.3% relative improvement, with LongMemEval_s* as the second benchmark.
Explicit logical structure at generation time: the retrieved rules give the answer generator a scaffold, which is what separates this from retrieval-only memory improvements.
Why it matters: the memory lane is crowded with storage-format papers. This one changes what memory is for, moving it from lookup to inference.
Abstract
Question answering agents in long-term conversations must reason over massive, temporally dispersed dialogue histories. However, existing memory mechanisms primarily treat past information as \textit{passively} stored facts, leading to semantic gaps and unreliable reasoning. To address this limitation, we propose RuleMem, a rule-based memory framework that induces reusable logical rules from historical interactions to \textit{actively} guide both evidence retrieval and reasoning. Specifically, RuleMem constructs natural-language Horn clauses from conversations and validates them via a Rule Perplexity Consistency (RPC) mechanism. These induced rules enable the retrieval of semantically distant evidence while providing an explicit logical structure for answer generation. We conducted a comprehensive evaluation of RuleMem on two long-term conversational benchmarks, LoCoMo and LongMemEval_s*. In a rigorous comparison against 14 baselines on LoCoMo, RuleMem achieved the highest accuracy, exceeding the baseline average by 27.47 points (a 54.3% relative improvement).