Researchers propose entity indexing to improve agent long-term memory

Xuanyu Meng and five coauthors submitted a preprint, EnSIMem, on September 23, proposing an agent long-term memory architecture organized by entity; the code is open-sourced on GitHub.

Existing memory systems often compress interactions into generic summaries or retrieve anonymous text chunks, making it hard for an agent to identify the right entity, property and evidence. EnSIMem organizes interactions offline into theme-coherent episodes and builds index entries of the form [entity][entity_type][property: value], each preserving source turns and temporal information; online, the request is decomposed into evidence requirements and answered from the index.

The accompanying repository shows the experiments run on the LoCoMo and LongMemEval long-term conversational memory benchmarks, with ablation scripts included. The abstract says the system achieves high answer accuracy while keeping contexts compact, but it names no baseline or number, so the performance claim awaits the paper body and independent reproduction.