基于节点兴趣的非结构化P2P网络的搜索机制

发布时间：2018-06-05 21:59

本文选题：聚类算法 + 节点兴趣　；参考：《新疆大学》2013年硕士论文

【摘要】：随着经济的发展，互联网的日益普及，网络资源和网民的数量呈爆炸式增长，必然会导致一些问题的出现，那就是在如此庞大的资源海洋中，如何快速准确全面的定位到自己想要的资源？面对越来越多的资源，传统的基于服务器的搜索引擎的弊端越来越明显，比如受服务器的计算能力和存储容量的影响，极大的影响了搜索的效率，P2P技术作为一种新兴的网络组织模式，为搜索引擎的发展提供了空间，各种搜索算法的提出，为提高搜索效率提供了帮助。与结构化P2P网络相比，非结构化P2P网络的节点缺乏全局的拓扑结构,随机的动态变化比较复杂，如何高效的搜索到用户感兴趣的资源是非结构化P2P网络搜索策略的关键问题。非结构化P2P网络最广泛的搜索策略是采用泛洪机制，但是随着网络节点以及资源数量的大规模增多，搜索的冗余消息也呈指数级增长，，严重影响了用户的搜索效率。关于此问题，本文提出了非结构化P2P网络的一种基于用户兴趣度的搜索策略，所包含的资源可以很好的表达出节点的兴趣，将节点按照其所表现出的兴趣组成网络，使兴趣度高的节点在网络中比较接近。用户在网络中检索感兴趣的资源，只需将查询消息路由到拥有相关资源的节点，从而提高检索效率。本文的主要内容有：（1）首先对P2P网络的概念和作用做了阐述，然后分别对结构化和非结构化P2P网络的特点及搜索算法做了详细的介绍，这些都是本论文的基础知识；（2）本文引入了基于知网的词汇相似度计算，详细介绍了知网以及词汇相似度计算的方法，在此基础上形成词集合，词汇相似度计算是本文算法的基础；（3）提出用k-高频集合向量来表达节点的兴趣，该方法是在k-高频词向量的基础上提出的，能够更好的表达出节点的兴趣；（4）提出了基于节点兴趣度的P2P网络的搜索算法，通过余弦算法计算节点间的兴趣相关度，把兴趣相关度高的节点建立关联，使得拥有同类资源的节点在同一个兴趣域内，要搜索此类资源，只需到相应的兴趣域查找，极大的提高的了搜索效率。随着搜索次数的增加，节点对邻居节点表不停的优化，使得搜索效率越来越高；（5）最后对本文提出的算法进行了仿真实验，通过和现有的基于节点兴趣的搜索算法对比表明：本文的算法大大减少了漏查的现象，提高了查找的全面率和搜索成功率。
[Abstract]:With the development of economy and the increasing popularity of the Internet, the number of network resources and Internet users is increasing explosively, which will inevitably lead to the emergence of some problems, that is, in such a vast ocean of resources, How to locate the resources you want quickly and accurately? In the face of more and more resources, the disadvantages of the traditional server-based search engine are becoming more and more obvious, such as the influence of the computing power and storage capacity of the server. As a new network organization mode, P2P technology has greatly affected the efficiency of search, and has provided space for the development of search engine. Compared with structured P2P networks, the nodes of unstructured P2P networks lack global topology, and the random dynamic changes are more complex. How to efficiently search the resources of interest to users is the key problem of unstructured P2P network search strategy. The most widely used search strategy in unstructured P2P networks is flooding mechanism, but with the increase of the number of network nodes and resources, the number of redundant messages increases exponentially, which seriously affects the search efficiency of users. To solve this problem, this paper proposes a search strategy based on user interest in unstructured P2P networks. The resources contained can express the interests of nodes and form the network according to their interests. The nodes with high degree of interest are closer to each other in the network. In order to improve the retrieval efficiency, users can only route the query message to the node that has the relevant resources in order to retrieve the interested resources in the network. The main contents of this paper are as follows: Firstly, the concept and function of P2P network are expounded, then the characteristics and search algorithm of structured and unstructured P2P network are introduced in detail. These are the basic knowledge of this paper. (2) this paper introduces vocabulary similarity calculation based on knowledge net, and introduces the method of word similarity calculation in detail. On the basis of this, the vocabulary similarity calculation is the foundation of this algorithm. This method is based on the k-high frequency word vector, which can better express the interest of the node. In this paper, a search algorithm of P2P network based on node interest is proposed. The correlation of interest between nodes is calculated by cosine algorithm, and the nodes with high degree of interest are associated, so that the nodes with similar resources are in the same domain of interest. In order to search such resources, we only need to search the corresponding domain of interest, which greatly improves the efficiency of the search. With the increase of search times, the nodes constantly optimize the neighbor table, which makes the search efficiency more and more high. Finally, the proposed algorithm is simulated and compared with the existing search algorithm based on nodal interest. It is shown that the algorithm greatly reduces the phenomenon of missing search, and improves the overall search rate and the search success rate.
【学位授予单位】：新疆大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP393.02

【相似文献】