Research on Parallel Data Currency Rule Algorithms

Abstract

Data currency is a temporal reference of data, which is related to the value of data and affects the results of data analysis and mining. The currency rules that reflect the time series features of data can be used not only for data repairing, but also for data quality evaluation. However, with the rapid growth and dynamic update of data volume, both the forms and algorithms of basic currency rule are facing severe challenges in application. Therefore, based on the research on data currency repairing, we extended the basic currency rule form, and proposed rule extraction and incremental updating algorithms that can run in parallel on dynamic data set. The experimental results show that, compared with non-parallel methods, the efficiency of parallel algorithms is significantly improved.