计算机科学
搜索引擎索引
正确性
数据挖掘
k-最近邻算法
调度(生产过程)
网格
Web查询分类
查询优化
吞吐量
分布式计算
情报检索
Web搜索查询
搜索引擎
算法
人工智能
电信
经济
数学
运营管理
几何学
无线
作者
Dan He,Sibo Wang,Xiaofang Zhou,Reynold Cheng
出处
期刊:IEEE Transactions on Knowledge and Data Engineering
[Institute of Electrical and Electronics Engineers]
日期:2019-09-20
卷期号:33 (4): 1554-1566
被引量:6
标识
DOI:10.1109/tkde.2019.2942585
摘要
The intelligent transportation systems, e.g., DiDi and Uber, have served as essential travel tools for customers, which foster plenty of studies for the location-based queries on road network. In particular, given a set O of objects and a query point q on a road network, the k Nearest Neighbor (kNN) query returns the k nearest objects in O with the shortest road network distance to q. In literature, most existing solutions for kNN queries tend to reduce the query time, indexing storage, or throughput of the kNN queries while overlooking the correctness of the queries caused by query-query and update-query conflicts. In our work, we propose a grid-based framework on conflict-aware kNN queries on moving objects which aims to optimize system throughput while guaranteeing query correctness. In particular, we first propose efficient index structures and new query algorithms that significantly improve the throughput. We further present novel scheduling algorithms that aim to avoid conflicts and improve the system throughput. Moreover, we devise approximate solutions that provide a controllable trade-off between the conflict of kNN queries and system throughput. Finally, we propose a cost-based dispatching strategy to assign the kNN results to the corresponding queries. Extensive experiments on real-world data demonstrate the effectiveness and efficiency of our proposed solutions over alternatives.
科研通智能强力驱动
Strongly Powered by AbleSci AI