藏文web网络社区划分研究与实现

发布时间：2018-01-02 23:15

本文关键词：藏文web网络社区划分研究与实现　出处：《西北民族大学》2016年硕士论文　论文类型：学位论文

【摘要】：大量的藏文Web页面中存在着海量关系,从底层观点来看,这些页面节点是由超链接互相指向而形成在一起的社区。本文首先对网页链接的处理技术做了介绍,分析了各种复杂网络模型,说明了模型的建立过程,对建立藏文Web网络模型中的难点进行了分析,包括识别一个GB码网页是否是藏文网页的问题,并采用高级语言设计了网页节点的数据结构。然后本文列出了社区划分中常用的三种网络模型:Karate club网络、Dolphin网络、Football网络,并与藏文Web网络模型一起做了分析。最后通过对模型产生背景的比较,从适用于已知社区个数和未知社区个数的不同种类算法不容易整合评价的角度,提出了用噪声节点去评价划分质量的方法。目前现存的社区划分算法大多针对无向图来进行叙述,而藏文Web网络中的页面节点和链接总体上是一种有向图结构,因此本文在分析以往经典的无向图社区划分算法之外,对各种理论上网络模型的生成过程进行了适合有向图的变化探讨,之后提出了一种基于星型网络拓扑的网页社区划分算法。将GN算法划出的社区个数控制在与本文算法划出个数相同的情况下,从噪声节点的角度对比本文算法,发现有不错的结果,并且最终采用可视化工具将其呈现出来。本文的创新点有：采用跨平台的Java语言实现的藏文Web识别,无标度网络的有向图社区划分算法,通过噪声节点去整合评价已知社区个数和未知社区个数的社区划分算法。在未来的进一步工作中,将会考虑对算法做适当的递归改写和多核优化。
[Abstract]:A large number of Tibetan Web pages have a large number of relationships, from the bottom of the point of view. These page nodes are a community formed by hyperlinks pointing to each other. Firstly, this paper introduces the processing technology of web links, analyzes various complex network models, and explains the establishment process of the models. The difficulties in establishing Tibetan Web network model are analyzed, including the problem of identifying whether a GB code web page is a Tibetan web page. The data structure of the web node is designed by using high-level language. Then, three common network models in community division are listed: the Dolphin club network and the Dolphin network. Football network and Tibetan Web network model are analyzed together. Finally, the background of the model is compared. From the point of view that different kinds of algorithms suitable for the number of known and unknown communities are not easy to integrate and evaluate. This paper proposes a method to evaluate the partition quality by noise-nodal nodes. Most of the existing community partitioning algorithms are described for undirected graphs. The page nodes and links in Tibetan Web network are a kind of directed graph structure in general, so this paper analyzes the classical undirected graph community partition algorithm. The generation process of various network models in theory is discussed in this paper, which is suitable for the change of directed graph. After that, a new algorithm of web page community partitioning based on star network topology is proposed, which controls the number of communities identified by the GN algorithm under the same number as the algorithm in this paper. From the point of view of noise-node comparison, we find that there are good results. Finally, the visualization tools are used to show it. The innovations of this paper are: Tibetan Web recognition using cross-platform Java language, non-scale-network directed graph community partition algorithm. The noise nodes are used to integrate the algorithm to evaluate the number of known and unknown communities. In the future work, appropriate recursive rewriting and multi-core optimization will be considered.
【学位授予单位】：西北民族大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP393.09

【相似文献】