SelRoute：基于查询类型的长期对话记忆检索路由
SelRoute: Query-Type-Aware Routing for Long-Term Conversational Memory Retrieval
AI总结：
SelRoute通过根据查询类型选择专用检索管道提升长期对话记忆检索效果，实验显示其在多个基准测试中表现优异，但存在推理密集型检索任务的失败模式。
摘要：
从长期对话记忆中检索相关过去交互通常依赖于大规模密集检索模型（110M-1.5B参数）或LLM增强索引。我们引入SelRoute，一种根据查询类型将每个查询路由到专用检索流程——词汇、语义、混合或词汇增强——的框架。在LongMemEval_M（Wu等，2024）上，SelRoute使用bge-base-en-v1.5（109M参数）达到Recall@5为0.800，使用bge-small-en-v1.5（33M参数）达到0.786，相比Contriever使用LLM生成的事实键达到0.762。一个零ML基准使用SQLite FTS5单独达到NDCG@5为0.692，已经超过了所有已发布的基准线在排序质量上的差距——我们部分归因于词汇检索实现差异。五折分层交叉验证确认了路由的稳定性（CV差距为1.3-2.4 Recall@5点；在折叠中4/6查询类型路由稳定）。基于正则表达式的查询类型分类器达到83%的有效路由准确率，而端到端检索使用预测类型（Recall@5 = 0.689）仍然优于统一基准。在8个额外基准测试上的交叉评估，涵盖62,000+实例——包括MSDialog、LoCoMo、QReCC和PerLTQA——证实了无基准特定调整的泛化能力，同时暴露了推理密集型检索任务的明显失败模式（RECOR Recall@5 = 0.149）限制了声明。我们还识别出一种增强-嵌入不对称性：存储时的词汇扩展提高了词汇搜索但损害了嵌入搜索，这促使了每条管道的增强决策。完整的系统在查询时不需要GPU和LLM推理。
英文摘要：
Retrieving relevant past interactions from long-term conversational memory typically relies on large dense retrieval models (110M-1.5B parameters) or LLM-augmented indexing. We introduce SelRoute, a framework that routes each query to a specialized retrieval pipeline -- lexical, semantic, hybrid, or vocabulary-enriched -- based on its query type. On LongMemEval_M (Wu et al., 2024), SelRoute achieves Recall@5 of 0.800 with bge-base-en-v1.5 (109M parameters) and 0.786 with bge-small-en-v1.5 (33M parameters), compared to 0.762 for Contriever with LLM-generated fact keys. A zero-ML baseline using SQLite FTS5 alone achieves NDCG@5 of 0.692, already exceeding all published baselines on ranking quality -- a gap we attribute partly to implementation differences in lexical retrieval. Five-fold stratified cross-validation confirms routing stability (CV gap of 1.3-2.4 Recall@5 points; routes stable for 4/6 query types across folds). A regex-based query-type classifier achieves 83% effective routing accuracy, and end-to-end retrieval with predicted types (Recall@5 = 0.689) still outperforms uniform baselines. Cross-benchmark evaluation on 8 additional benchmarks spanning 62,000+ instances -- including MSDialog, LoCoMo, QReCC, and PerLTQA -- confirms generalization without benchmark-specific tuning, while exposing a clear failure mode on reasoning-intensive retrieval (RECOR Recall@5 = 0.149) that bounds the claim. We also identify an enrichment-embedding asymmetry: vocabulary expansion at storage time improves lexical search but degrades embedding search, motivating per-pipeline enrichment decisions. The full system requires no GPU and no LLM inference at query time.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
