Zhang Yan, Wang Hongzhi, Yang Zhongsheng, Li Jianzhong
Department of Computer Science and Technology, Harbin Institute of Technology, Harbin, China.
PLoS One. 2014 Aug 18;9(8):e103853. doi: 10.1371/journal.pone.0103853. eCollection 2014.
The quality of data plays an important role in business analysis and decision making, and data accuracy is an important aspect in data quality. Thus one necessary task for data quality management is to evaluate the accuracy of the data. And in order to solve the problem that the accuracy of the whole data set is low while a useful part may be high, it is also necessary to evaluate the accuracy of the query results, called relative accuracy. However, as far as we know, neither measure nor effective methods for the accuracy evaluation methods are proposed. Motivated by this, for relative accuracy evaluation, we propose a systematic method. We design a relative accuracy evaluation framework for relational databases based on a new metric to measure the accuracy using statistics. We apply the methods to evaluate the precision and recall of basic queries, which show the result's relative accuracy. We also propose the method to handle data update and to improve accuracy evaluation using functional dependencies. Extensive experimental results show the effectiveness and efficiency of our proposed framework and algorithms.
数据质量在业务分析和决策中起着重要作用，而数据准确性是数据质量的一个重要方面。因此，数据质量管理的一项必要任务是评估数据的准确性。为了解决整个数据集准确性低而其中有用部分准确性可能高的问题，还需要评估查询结果的准确性，即相对准确性。然而，据我们所知，尚未提出用于准确性评估方法的度量标准或有效方法。受此启发，对于相对准确性评估，我们提出了一种系统方法。我们基于一种使用统计量来度量准确性的新指标，为关系数据库设计了一个相对准确性评估框架。我们应用这些方法来评估基本查询的精确率和召回率，以显示结果的相对准确性。我们还提出了处理数据更新以及使用函数依赖来改进准确性评估的方法。大量实验结果表明了我们所提出的框架和算法的有效性和效率。
