Deep Web集成系统中同类主题数据源选择方法

摘要

This paper presented a similar theme of Deep Web data sources selection which could effectively know repeatability of content between new data source and integrated system by differences analysis of the data source, then used precision and recall to construct a quality estimation model for assessing quality of each data source, weakened negative impact of quality assessment that dues to low precision in existing research. Experiment results which use mainstream book sites show that this method can reduce the burden of the system, and obtain higher quality from data sources with similar theme.%提出了一个同类主题的Deep Web数据源选择方法,该方法通过数据源差异性分析可有效判断出新数据源的内容与集成系统中已有内容的重复度,进而利用查准率和查全率建立质量估计模型评估各数据源的质量,削弱了已有研究中因查准率低对质量评估产生的负面影响.在主流图书类网站上的实验结果表明,该方法能减少系统的负担,同时获取质量较高的同类主题的数据源.