基于列重合度的网络表格一致性扩展

发布时间：2018-02-27 15:42

本文关键词： 网络表格扩展列重合度列映射一致性支持度　出处：《计算机科学》2017年09期 　论文类型：期刊论文

【摘要】：网络表格的扩展是根据已知信息扩展与主列相关的其他属性列,以满足人们通过表格获取感兴趣信息的需求。目前的研究工作主要针对由主列和待扩展列组成的实体-属性二元表,并将主列视为其他属性列扩展的唯一依据,但该技术运用到具有多个待扩展列的网络表格时,由多个二元表拼接而成的结果表很容易出现实体不一致现象。综合考虑各属性列间以及元组行间的关系,提出一致性支持度概念,设计并实现了基于列重合度的表格一致性扩展系统CCA,其既能保证候选值的高匹配分数,又能使结果表中填值所使用的数据源表数目最小化,有效地避免了实体不一致问题。实验表明,与现有方法相比CCA系统有更高的精确度、覆盖率、一致性,以及更低的查询时间代价。
[Abstract]:The extension of the network table extends the other attribute columns associated with the main column based on known information, In order to meet the need of obtaining information of interest through tables, the current research mainly focuses on the entity attribute binary table, which consists of the main column and the column to be expanded, and regards the main column as the sole basis for the extension of other attribute columns. However, when this technique is applied to network tables with multiple columns to be expanded, the result tables, which are composed of multiple binary tables, are prone to entity inconsistency. The relationships between attribute columns and tuple rows are considered synthetically. The concept of consistency support is proposed, and a table consistency extension system (CCAA) based on column coincidence degree is designed and implemented. It can not only guarantee the high matching score of candidate values, but also minimize the number of data source tables used for filling out values in the result table. Experimental results show that CCA systems have higher accuracy, coverage, consistency and lower query time cost than the existing methods.
【作者单位】：北京交通大学计算机与信息技术学院;
【基金】：国家自然科学基金面上项目(61370060)资助
【分类号】：TP393.0

【相似文献】