因果知识发现在关系数据集中的应用研究

因果发现是知识发现的一个新的研究方向,其目的是从观测数据中找出其隐含的因果关系。虽然应用前景十分美好,但结果评价困难等因素造成了其在应用方面研究的缺乏,并限制了它的进一步发展。文章对于一个人口统计方面的关系型数据集进行了因果发现,并对于结果进行了多方面的评估与分析。结果表明,因果发现不但切实可行,而且能在实际的知识发现应用中找出相对来说较高质量的知识。

Application of Causal Discovery to Relational Dataset

Causal discovery is a new research direction in KDD field.Its aim is to find the causal relations in the ob-servational datasets.Despite of its promising future,the lack of successful applications due to some factors such as the difficulty of the evaluation prevents its farther development.This paper processes a real census relational dataset with a causal discovery algorithm.After that,it evaluates and analyzes the results from several aspects.The evaluation and analy-sis prove that the causal discovery is feasible,and it can find knowledge of good quality in the real applications.