【摘要】 针对当前冲突检测算法中存在的大量顺序比较运算导致冲突检测性能不高的问题,提出采用决策树模型的分类思想来提高策略冲突检测效率的方法。首先,为实现对规则中存在的复合型维度进行分类,采用分解算法分解复合型维度为集合;其次,为减少策略规则之间的顺序比较,通过切割规则中的各个维度来构造决策树,实现对规则的分类;最后仅对决策树叶子节点中的规则进行顺序比较来检测所有规则之间可能存在的冲突。通过对规则进行分类,把可能存在冲突的规则切割到同一个叶子节点中,从而减少了不必要的顺序比较。实验结果表明,当策略规则达到2 000条时,冲突检测速度提高20倍以上,且规则越多,检测速度提升越明显。另外,冲突检测采用顺序比较,能保证冲突检测的准确性为100%。更多还原

【Abstract】 For the poor efficiency in the policy conflict detection algorithm caused by large scale sequence comparison,decision tree model was used to improve the detection efficiency of policy conflict.Firstly,in order to realize classification on the complex dimension of rules,a decomposition algorithm was applied to decompose the complex dimension;secondly,for the purpose of reducing the sequence comparisons between policy rules,rules were cut according to the various dimensions and a decision tree was constructed;finally,only by comparing rules in leaf node of decision tree all the conflicts could be detected.Through the classification for rules,the rules which may contain conflict were cut to the same leaf node,reducing the unnecessary sequence comparison.Experimental results show that the efficiency of conflict detection increase by more than20 times when 2 000 policy rules are contained,and the more rules,the more efficient detection.In addition,sequence comparison was employed in the conflict detection,which ensured the accuracy of conflict detection of 100%.更多还原