基于函数依赖与条件约束的数据修复方法

发布时间：2018-07-23 09:20

【摘要】：随着经济与信息技术的发展,在许多应用中均产生大量数据.然而,受硬件设备、人工操作、多源数据集成等诸多因素的影响,在这些应用之中往往存在较为严重的数据质量问题,特别是不一致性问题,从而无法有效管理数据.因此,首要的任务就是开发新型数据清洗技术来提升数据质量,以支持后续的数据管理与分析.现有工作主要研究基于函数依赖的数据修复技术,即以函数依赖来描述数据一致性约束,通过变更数据库中部分元组的属性值(而非增加/删除元组)来使得整个数据库遵循函数依赖集合.从一致性约束描述的角度来看,函数依赖并非是唯一的表达方式,还存在其他表达方式,例如硬约束、数量约束、等值约束、非等值约束等.然而,随着一致性约束种类的增加,其处理难度也远比仅有函数依赖的场景要困难.考虑以函数依赖与其他一致性约束共同表述数据库的一致性约束,并在此基础上设计数据修复算法,从而提升数据质量.实验结果表明,所提方法的执行效率较高.
[Abstract]:With the development of economy and information technology, a lot of data are produced in many applications. However, due to many factors such as hardware, manual operation and multi-source data integration, there are many serious data quality problems in these applications, especially inconsistency, which makes it impossible to manage data effectively. Therefore, the first task is to develop new data cleaning technology to improve data quality to support subsequent data management and analysis. The existing work mainly studies the data repair technology based on functional dependency, that is, describing data consistency constraints by functional dependency. By changing the attribute values of some tuples in the database (rather than increasing / deleting tuples), the entire database follows the set of functional dependencies. From the point of view of consistent constraint description, functional dependency is not the only way of expression, but there are other expressions, such as hard constraint, quantity constraint, equivalent constraint, non-equivalent constraint and so on. However, with the increase of the types of consistency constraints, the processing difficulty is much more difficult than that of only functional dependency scenarios. In order to improve the quality of data, we consider using function dependency and other conformance constraints to express the consistency constraints of the database, and then design a data repair algorithm based on this. The experimental results show that the proposed method is more efficient.
【作者单位】：华东师范大学计算机科学与软件工程学院数据科学与工程研究院;
【基金】：国家重点基础研究发展计划(973)(2012CB316203) 国家自然科学基金(61370101,U1501252,61532021) 上海市教委科研创新重点项目(14ZZ045)~~
【分类号】：TP311.13

【参考文献】