不完整数据高效近似查询处理算法

Efficient Approximate Query Processing Algorithm for Incomplete Data

当前的不完整数据查询处理算法没有将冗余数据和脏数据清洗,而且寻优过程缓慢,不利于数据查询结果的快速展示。提出将各数据阅读器和各局部过滤器连接,利用局部过滤器对数据阅读器所传输的脏数据和多读数据进行一次局部性地过滤,再由各个局部过滤器把初步清洗的数据发送到全局过滤器,且由全局过滤器依据阅读器空间位置以及其他信息,实现包含添加漏读数据和删除多读数据以及冗余数据的进一步清洗,以提高查询效率。将Rank List结构作为索引,利用Topk数据结构有序性的特点,对不完整的数据合理利用,高效查询到前K个非常有代表性的Skyline点,将查询结果展示出来。通过实验证明,所提算法有效地过滤了冗余数据,提高了查询处理的效率,可行性较高。

The current incomplete data query processing algorithm does not clean the redundant data and dirty data,and the optimization process is slow,which is not conducive to the rapid display of data query results.It is proposed to connect each data reader and each local filter,and use the local filter to filter the dirty data and the multi-read data transmitted by the data reader once,and then send the preliminary cleaning data to each local filter by each local filter Global filter,and by the global filter based on the reader space location and other information,including the addition of missing read data and delete multiple read data and redundant data to further cleaning to improve query efficiency.The Rank List structure as an index,the use of Topk data structure of the characteristics of orderly,the use of incomplete data,efficient query to the former K very representative Skyline point,the query results displayed.Experiments show that the proposed algorithm effectively filters the redundant data and improves the efficiency and feasibility of the query processing.