摘要
针对现有网络数据划分算法对特性体现不充分的问题,提出一种新的基于网络数据中心度的聚类划分算法。该算法提出了考虑属性相同数据关系强度的数据集中心度,应用到Louvain算法的第二阶段作为到种子节点选取的依据,促使小网络数据集合并至大网络数据集,同时抑制大网络数据间的过度合并。本文以真实网络数据集进行验证,选取多种评价指标定量评估划分结果 ,并采用布局算法对网络数据划分结构做可视化展示。实验表明本文提出的网络数据划分算法能更好地挖掘数据集合,划分结果也更加合理。
In this paper, a new clustering partitioning algorithm based on network data centrality is proposed. The algorithm considers the attributes of the same data relationship strength data of heart, as to the seed node selection applies to the second phase of the Louvain algorithm on the basis of data collection and network, makes small to large network data sets, and inhibits the excessive merger between data network. This paper validates the real network data sets, selects a variety of evaluation indicators, quantitatively evaluates the segmentation results, and uses the layout algorithm to visualize the network data partition structure. Experiments show that the proposed network data partitioning algorithm can better mine data sets, and the results are more reasonable.
作者
李贤
许大卫
LI Xian;XU Da-wei(State Grid Jibei Information ＆ Telecommunication Company,Beijing 100053 China)
出处
《自动化技术与应用》
2018年第9期86-90,103,共6页
Techniques of Automation and Applications
