【作者】 王小雨；

【导师】 刘建辉；

【作者基本信息】 辽宁工程技术大学 ， 计算机应用技术， 2011， 硕士

【摘要】 复杂网络社团结构的研究为人类发现更多实际意义的社团提供了更多的借鉴。本文主要研究了复杂网络的社团结构探测算法,通过对已有算法的学习和研究,改进了一种基于K-means的算法,在不知道社团结构的前提下对复杂网络进行划分,算法简单、易理解,把算法应用在karate网络中,实验结果表明此算法是有效的。另外借鉴基于节点密度等性质,提出了一种基于节点间相似度的复杂网络社团结构探测算法(BSTN),此算法迭代次数大大减少,在计算机生成的已知社团结构的随机网络中检验,结果表明此算法比GN算法具有更高的准确率。另外还在实际网络中进行验证,本文使用的是空手道俱乐部网络(karate网络)、美国大学足球俱乐部网络(Football网络)和电子邮件网络(Email网络),实验结果与Newman算法进行了比较,本文提出的算法拥有更少的迭代次数,近似或者更大的模块化函数值,说明此算是有效的;并且能够对算法得出的社团结构进行了合理地解释,说明算法划分得出的结果是符合实际的,是合理的。更多还原

【Abstract】 The research on complex network community structure has provided more references for people find more practical significance of the communities. This paper mainly studies the complex network detection algorithm, through learning and research on existing algorithms, and improves an algorithm based on K-means, which detects the community structure of complex networks under the premise of unknown community structure. The algorithm is simple, easy understanding. Using the algorithm in network, the experimental results show that this algorithm is effective. Another reference node density properties, this paper puts forward a method community structure detection algorithms (BSTN) based on similarity between the nodes of the complex network, the algorithm greatly reduce iteration times, using the algorithm in the computer generated stochastic network known community structure, the result shows that this algorithm has higher accuracy than GN algorithm. Also in the actual network, this paper uses karate club network (karate network), the American College Football club network (football network) and the Email network, experimental results compare to Newman algorithm, the proposed algorithm can have less iteration, approximate or more larger value of the module, it shows the BSTN algorithm is effective, and reasonable explain the community structure getting from the BSTN algorithm, the result of from the BSTN algorithm is practical, is reasonable.更多还原