向量数据库系统近似最近邻搜索综合实证评估
原标题：A Comprehensive Empirical Evaluation of Vector Database Systems for Approximate Nearest Neighbor Search: Performance, Quality, and Resource Trade-offs
AI 摘要
该研究对七种主流向量数据库系统（FAISS、Qdrant、Milvus、Weaviate、Chroma、pgvector、LanceDB）进行了全面实证评估，覆盖六个数据集、超过400万向量，测量了检索质量、查询性能和资源消耗等15项指标。结果显示FAISS吞吐量最高但缺乏数据库功能，Weaviate召回率最佳，Qdrant延迟最低，LanceDB索引构建更快但检索质量较低。研究提供了系统选择指南，并开源了基准测试框架。
正文节选
A Comprehensive Empirical Evaluation of Vector Database Systems for Approximate Nearest Neighbor Search: Performance, Quality, and Resource Trade-offs Abstract Vector databases have emerged as critical infrastructure for modern artificial intelligence applications, particularly retrieval-augmented generation (RAG), semantic search, and recommendation systems. Despite their growing importance, there remains a significant gap in comprehensive, reproducible benchmarks that jointly evaluate retrieva