摘要
建立统一的独立社区和重叠社区挖掘结果评价标准,对于简化复杂网络社区挖掘、提高结果评估可比性有重要意义.目前的独立社区评价没有解决社区划分何时中止的问题,而重叠社区评价标准则往往具有较大的主观性.本文基于链接关联度和最优节点复制下的节点关联度,提出了一种支持独立和重叠社区挖掘、客观的统一评价模型,且基于该评价模型的社区挖掘方法可以进行多尺度的独立社区和重叠社区挖掘.理论分析和实际网络实验证实了该评价模型的合理性和可用性.
A uniformed evaluation criterion of independent or overlapping community detection is of great significance for simplifying detecting and improving result comparability. The existing approaches for independent community detection can＇t decide how many communities should be discovered, while those for overlapping community detection are somewhat subjectively proposed. In this paper a novel objective evaluation model is proposed, which is based on the Correlation coefficient of links and the optimized duplication of nodes ,detection algorithm based on the model is convenient and supports muM-scale community discovery. Theoretical analysis and experiments indicate the model＇s reasonability and usability.
出处
《小型微型计算机系统》
CSCD
北大核心
2016年第11期2428-2432,共5页
Journal of Chinese Computer Systems
基金
国家自然科学基金项目(61262014)资助
江西省自然科学基金项目(20132BAB201034)资助
江西省教育厅科技项目(GJJ13224)资助
江西师范大学博士启动基金
关键词
社区挖掘
统一评价模型
关联度
最优节点复制
复杂网络
community detection
uniformed evaluation model
correlation coefficient
optimized node duplication
complex network
