最小权重有向频繁子图挖掘

Mining in minimum weighted frequent directed subgraphs

权重有向图作为众多网络结构的抽象模型,是当前数据挖掘研究的热点,可作为厂区铁路线路的模型。本文针对权重有向图进行挖掘,提出了两种新算法,可以得到连通完整的子图。第1种算法WDSpan采用深度优先搜索策略在完成完整挖掘的基础上比较权重值的大小,第二种算法MWD以不同子图规模的平均权重和支持度之积作为新的计算度量,在挖掘过程中考虑权重因素,并在满足条件的子图中找到不同图规模的最小权重子图,实验证明该算法节省了存储空间。

Weighted directed graph, being the abstract model of many network structures, was the hotspot of data mining research currently, which could be used for the model of factory railway line as well. This paper aimed at mining weighted directed graph, two kinds of new algorithms were proposed and a connected complete subgraph could be obtained. The first algorithm WDSpan was used to compare the value of weight on the basis of using depth-first search strategy to complete integral mining. The second algorithm called MWD was used to take the product of different subgraph sizes＇ average weights and support line as a new calculation measure, the factor of weight was considered in the process of mining and the minimum weight subgraphs of different models were in subgraphs that met the condition. Experimental results showed that the algorithm reduced the space of memory.