多关系数据挖掘是借鉴ILP技术,并结合机器学习方法所提出的数据挖掘新课题。多关系关联规则是多关系方法在概念描述任务中最具代表性的研究方向之一,此类方法在发挥多关系方法的模式表达能力与利用背景知识能力的同时.借鉴成熟的关联规则方法的思想与优化策略,取得了较高的性能与表达复杂模式的能力,同时在面向复杂结构数据的应用中获得了较好的效果。在简述多关系方法的基础上,通过分析与比较目前具有代表性的多关系关联规则算法,总结了各算法的优势与不足,并指出了该领域目前的主要热点问题。
Multi-relational data mining,which is inspired by ILP and combined with methods of Machine Learning,is a new topic in data minning.Mining multi-relational association rules is one of the most representative fields of multi-relational data mining in the task of concept description.These approaches are provided with powerful expressiveness and the ability to exploit the background knowledge,and adopt the ideas and strategy of association algorithms.In this paper,the concept of multi-relational data mining is introduced firstly,then the nowadays typical algorithms in this field are analyzed and compared,afterwards the summaries of features of those algorithms are presented,finally the relatively focused issues are pointed.