Temporal Spatial-Keyword Search on Databases Using SQL
Massive amount of textual content is associated with location and time tags,generated on webs related to restaurant,group-buying or social networking services.Often,users tend to retrieve up-to-date information with location and text proximity to some specified descriptions.To accelerate the search process,the state-of-the-art methods resort to design new index structures and probing algorithms.Nevertheless,efficient solutions fully supported by existing RDBMS still remain an open problem.To address this problem practically,in this paper,we propose TSKSQL,a solution that processes temporal spatial-keyword similarity search using SQL statements only.The novelty and advantages of TSKSQL are listed below.(1) We design a novel signature generation scheme that is able to properly capture texture,locational and temporal information properly.We index objects based on their generated signatures using a single B+-Tree with the ability to process similarity queries by simply probing the B+-Tree.(2) We propose various optimization techniques based on RDBMS so that both CPU and I/O costs can be reduced significantly.(3) We deploy TSKSQL over a real RDBMS,PostgreSQL.We conduct extensive experiments and the results show that TSKSQL demonstrates a good efficiency and stability.
Jingru Wang Jiajia Hou Feiran Huang Wei Lu Xiaoyong Du
DEKE,MOE and School of Information,Renmin University of China,Beijing 100872,China
国际会议
International Asia-Pacific Web Conference(第18届国际亚太互联网大会)
苏州
英文
204-216
2016-09-23（万方平台首次上网日期，不代表论文的发表时间）