Cooperative Memory Paging with Keyword Bookmarks for Long-Horizon LLM Conversations

TL;DR

This paper introduces cooperative paging with keyword bookmarks and a recall tool to enhance long-horizon LLM conversations, outperforming existing methods in answer quality on the LoCoMo benchmark.

Contribution

It proposes a novel cooperative paging method with keyword bookmarks and evaluates various design choices, significantly improving retrieval accuracy and conversation quality.

Findings

Coarse fixed-size pages achieve 96.7% coverage; content-aware topic shifts drop to 56.7%.

Eviction policy effectiveness varies with data; FIFO works best on synthetic data, LFU on LoCoMo.

Bookmark generation strategies improve performance; keyword specificity crucial for correct page selection.

Abstract

When LLM conversations grow beyond the context window, old content must be evicted -- but how does the model recover it when needed? We propose cooperative paging: evicted segments are replaced with minimal keyword bookmarks ([pN:keywords], ~8-24 tokens each), and the model is given a recall() tool to retrieve full content on demand. On the LoCoMo benchmark (10 real multi-session conversations, 300+ turns), cooperative paging achieves the highest answer quality among six methods -- outperforming truncation, BM25, word-overlap retrieval, a search-tool baseline, and full context -- on four models (GPT-4o-mini, DeepSeek-v3.2, Claude Haiku, GLM-5), confirmed by four independent LLM judges ($p=0.017$, paired bootstrap). We then study the paging design space with a 5x4 ablation over boundary strategies and eviction policies (3,176 synthetic probes, 1,600 LoCoMo probes). Key findings: (1) coarse fixed-size pages (fixed_20) reach 96.7% while content-aware topic_shift collapses to 56.7%; (2) eviction policy choice is data-dependent (FIFO best on synthetic, LFU on LoCoMo); (3) two bookmark generation strategies improve over the heuristic baseline (+4.4 and +8.7 E2E points); (4) the remaining bottleneck is bookmark discrimination -- the model triggers recall() 96% of the time but selects the correct page only 57% when bookmarks are insufficiently distinctive. Keyword specificity alone accounts for a 25 percentage point accuracy difference.