CueMem improves long-term chat memory by reconstructing dialogue context
CueMem: Cue-Guided Context Reconstruction for Long-Term Conversational Memory
Computation and Language
Summary
When chatbots try to remember long conversations, keeping all the chat history can be slow and confusing. The authors created CueMem, a system that stores small memory clues linked to the original chat parts. When asked a question, CueMem finds these clues and rebuilds just the important parts of the conversation to answer better. This method is faster and more accurate than using the entire chat history or compressed summaries.
What this means in practice
- •For chatbot developers: Build chatbots that better answer questions from long conversations by reconstructing relevant dialogue context from linked memory cues.
- •For customer service platform teams: Enhance virtual assistants to recall important past dialogue efficiently and accurately for improved user support experiences.
Authors
Changjian Wang, Rongzhen Li, Weili Guan, Shuming Shi, Quan Lu, Ning Jiang
Abstract
Long-term conversational agents must answer user queries by recalling information from extended dialogue histories, yet directly using the full history is costly and often unreliable, while compressed memory units may lose fine-grained evidence needed for question answering. Motivated by the reconstructive view of autobiographical memory, we propose CueMem, a cue-guided framework that treats extracted memory records as retrieval cues rather than self-contained evidence and reconstructs query-relevant dialogue context from their source turns. During memory construction, CueMem extracts fine-grained memory cues from dialogue turns and links each cue to its source turn. At query time, it retrieves query-relevant cues, maps them to source-turn anchors, and expands from these anchors over a turn graph that captures temporal proximity and semantic relatedness, reconstructing a compact evidence context from the original dialogue for LLM answer generation. Experiments on LoCoMo and LongMemEval show that CueMem consistently outperforms representative long-term memory baselines. Further analyses show that graph-based context reconstruction helps recover supporting dialogue evidence while reducing query-time input tokens and latency compared with the full-history LLM setting. These results highlight retrieval cues as an effective alternative to self-contained memory evidence for long-term conversational question answering.