数据库自适应查询优化技术研究  

Research on database adaptive query optimization technology

作　　者：魏远征 WEI Yuanzheng(Henan Technician College of Medicine and Health,Kaifeng,Henan 475000,China)

机构地区：[1]河南医药健康技师学院,河南开封475000

出　　处：《计算机应用文摘》2023年第16期77-79,共3页Chinese Journal of Computer Application

摘　　要：近似最近邻搜索是信息检索领域的基本技术之一,它能在大规模数据集上以较低的内存占用和更快的查询速度实现近似查询。其中,基于图的近似最近邻搜索算法是最常用的方法,该算法能够构建高质量的图索引结构,实现快速而准确的查询。然而,现有算法中存在一些查询点效率低的问题。为解决这一问题,文章提出了一种自适应查询方法,该方法将中间结果和查询向量作为模型训练的特征向量对梯度提升决策树模型进行特征训练,并将其整合到图索引结构上,以实现快速而准确的查询。实验结果表明,相对于基准算法,文章算法平均查询时间最长可减少64.4%,取得了显著的性能提升。Approximate nearest neighbor search is one of the basic technologies in the field of information retrieval,which can achieve approximate queries on large-scale datasets with lower memory usage and faster query speed.Among them,the graph based approximate nearest neighbor search algorithm is the most commonly used method,which can construct high-quality graph index structures and achieve fast and accurate queries.However,there are some issues with low query point efficiency in existing algorithms.In order to solve this problem,this paper proposes an adaptive query method,which uses the intermediate results and query vectors as the feature vectors of model training to train the features of the gradient lifting decision tree model model,and integrates them into the graph index structure to achieve fast and accurate query.The experimental results show that compared to the benchmark algorithm,the article algorithm can reduce the average query time by 64.4%,achieving significant performance improvement.

关 键 词：自适应优化 GBDT Link and Code数据库 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]