数据质量多种性质的关联关系研究

Association Relationships Study of Multi-Dimensional Data Quality

信息化时代数据海量增长的同时,用户需要利用多种指标从不同性质角度对数据质量进行评价和改善.但在目前数据质量管理过程中,影响数据可用性的多种重要因素并非完全孤立,在评估机制和指导数据清洗规则时,彼此会发生关联.研究了在实际信息系统中适用的综合性数据质量评估方法,将文献所提出以及在实际的信息系统中常用的数据质量性质指标按其定义与性质进行了归纳总结,提出了基于性质的数据质量综合评估框架.之后针对影响数据可用性的4个重要性质:精确性、完整性、一致性以及时效性整理出在数据集合上的操作方法,并逐一介绍其违反模式的定义,随后给出其具体关系证明,进而确定数据质量多维关联关系评估策略,并通过实验验证了该策略的有效性.

Recently, with the rapid growth of data quantity, users are using a variety of indicators to evaluate and improve the quality of data from different dimensions. During the course of data quality management, it is found that many important factors that influence the data availability are not completely isolated. In the evaluation mechanism which can guide data cleaning rules, these dimensions may be associated with each other. In this paper, serveral data quality dimensions researched in the literature as well as being used in the real information system are discussed, and accordingly the definition and properties of the dimensions are summarized. In addition, a multi-dimensional data quality assessment framework is proposed. According to the four important properties of data availability： Accuracy, completeness, consistency and currency, the operation method and the relationships among them on the data set are constructed. Finally, a multi-dimensional data quality accessment strategy is created. The effctiveness of the proposed strategy is verified by experiments.