信任网络中多维信任序列挖掘与安全防御研究

发布时间：2018-06-06 19:33

本文选题：信任网络 + 多维序列模式　；参考：《浙江工业大学》2014年硕士论文

【摘要】：近年来,开放网络环境中的诸多应用如P2P共享、电子商务、社交服务等已深刻地改变了人们的社会活动和交流方式,然而,在这种开放的、分布式的虚拟社会网络中存在着大量的欺诈行为和不可靠服务,严重影响了用户节点间的交互质量,安全问题已经成为制约网络应用发展的关键因素。信任网络能有效降低用户间的交互风险,提高系统服务质量和可靠性,已成为国内外研究的热点。然而,在潜在的巨大利益驱使下,针对信任网络的各种恶意攻击行为导致用户之间无法建立可靠的信任关系,极大地影响了信任网络的安全性和信任评价机制的有效发挥。现有信任研究大多侧重于信任关系计算与信任推理这两个方面,缺乏对信任网络中信任实体重要性及其关联关系的讨论,同时还忽略了从网络结构重要性角度考虑信任网络的安全问题,针对信任网络的安全保护研究还未引起人们足够的重视。因此,本文将提出多维序列模式挖掘方法分析信任网络中节点重要性及其关联结构,并深入研究和分析处于网络中不同地位的节点对信任网络安全的影响和防御恶意攻击的作用。本文主要贡献如下： 1.提出一种多维信任序列模式挖掘算法MTrustSeq,该算法包括频繁信任序列挖掘和多维模式筛选两个处理过程,综合考虑信任强度、路径长度和实体可信度等多维度因素,有效地挖掘出多维信任序列模式所包含的重要实体及其关联结构。仿真实验表明本文所提算法的挖掘结果全面、准确地反映了信任网络中重要实体及其关联结构特征。 2.考虑节点重要性对安全防御的影响,结合网络结构特性提出一种节点层次化划分方法levelDiv,分析节点在影响信任传递和网络整体安全方面的重要性,按重要性高低将信任网络结构划分成4个类别：核心节点集、重要节点集、关联节点集和无关节点集。实验表明该方法克服了传统的节点度、介数等度量指标的缺陷,能有效提高信任网络在恶意节点检测和安全防御方面的能力。
[Abstract]:In recent years, many applications in the open network environment, such as P2P sharing, e-commerce, social services and so on, have profoundly changed people's social activities and the way of communication, however, in this kind of open, There are a lot of fraud and unreliable services in distributed virtual social networks, which seriously affect the quality of interaction between user nodes. Security has become a key factor restricting the development of network applications. Trust networks can effectively reduce the risk of interaction between users, improve the quality of service and reliability of systems, and have become a hot research topic at home and abroad. However, all kinds of malicious attacks against trust networks, driven by potentially enormous benefits, make it impossible to establish reliable trust relationships between users. This has greatly affected the security of trust networks and the effective use of trust evaluation mechanisms. Most of the existing trust studies focus on trust relationship calculation and trust reasoning. There is a lack of discussion on the importance of trust entities and their relationships in trust networks, and the security of trust networks is neglected from the point of view of network structure importance. The research on the security protection of trust networks has not attracted enough attention. Therefore, this paper proposes a multi-dimensional sequential pattern mining method to analyze the importance and association structure of nodes in trust networks. The influence of nodes in different positions on the security of trust network and the role of defending against malicious attacks are studied and analyzed. The main contributions of this paper are as follows: 1. A multi-dimensional trust sequence pattern mining algorithm MTrustSeqis proposed. The algorithm includes two processes: frequent trust sequence mining and multi-dimension pattern filtering. The algorithm considers the multi-dimensional factors such as trust intensity, path length and entity credibility. The important entities and their association structures contained in the multidimensional trust sequence pattern are effectively mined. Simulation results show that the mining results of the proposed algorithm are comprehensive and accurately reflect the features of important entities and their associated structures in trust networks. 2. Considering the influence of node importance on security defense, a hierarchical node partition method, level Div, is proposed based on the characteristics of network structure, and the importance of nodes in influencing trust transfer and overall network security is analyzed. According to the importance, the trust network structure is divided into four categories: core node set, important node set, association node set and independent node set. Experiments show that the proposed method overcomes the shortcomings of traditional metrics such as node size and intermediate number and can effectively improve the ability of trust network in malicious node detection and security defense.
【学位授予单位】：浙江工业大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.08

【参考文献】