社会网络中的链路预测及网络重构

发布时间：2018-05-10 12:15

本文选题：社会网络 + 链路预测　；参考：《东北大学》2014年硕士论文

【摘要】：社会网络中人物之间关系的发掘一直是该研究领域中最重要的问题。互联网的普及极大的推动了在线社会网络的发展,这使得对社会网络的研究日益趋于复杂化。本文采用复杂网络研究方法对社会网络复杂性进行分析,并采用链路预测的思想实现对社会关系网络中关系的发现。另外,针对社会网络探测时因探测手段产生的关系抽取不全或者关系抽取错误的情况,提出了使用链路预测算法实现对探测到的社会网络重新构建的方案。本文从链路预测的角度出发研究社会网络中节点间产生连接关系的规律,分别对比分析了多种典型的链路预测算法在社会网络中的预测效果。针对现有的基于随机游走的链路预测算法存在的不足提出了基于邻居修正的随机游走链路预测算法。新提出的算法解决了原来随机游走算法可能出现的邻居选择单一的情况,由于考虑了源节点的所有邻居节点对最终相似度的贡献,因此对原来节点间的相似度进行了修正。经过邻居修正过后的随机游走链路预测算法能够取得很好的预测效果,并且该算法拥有较低的时间复杂度和较高的算法稳定性。将该算法应用到社会网络的好友的推荐上时获得了较为理想的推荐效果。另一方面,本文从社会人际关系网络重构的角度出发,分析并实现对探测到的人际关系网络的重新构建。针对面向中文文本的人际关系网络探测方法中存在的不足,改善并得到了能够很好揭示文本中人物之间关系的模型。以《三国志》文本为研究对象,通过人物抽象模型得到三国时期人物之间的关系网络,而后使用基于邻居修正的链路预测算法对其关系预测,并根据预测的结果实现对探测到网络的重新构建。发现,重构后的网络能够更真实的揭示三国时期的人物之间的关系。本文提出的基于邻居修正的随机游走链路预测方法有助于人们深入了解社会网络演化机制及人物之间产生联系的驱动力,实现对社会网络中好友的推荐。网络重构的思想则对真实社会关系网络的构建具有重要指导意义。
[Abstract]:The discovery of the relationship between people in social networks has been the most important issue in this field. The popularity of the Internet has greatly promoted the development of online social networks, which makes the research of social networks more and more complicated. In this paper, the complexity of social network is analyzed by using the method of complex network research, and the relationship in social network is discovered by the idea of link prediction. In addition, a link prediction algorithm is proposed to reconstruct the detected social network. From the point of view of link prediction, this paper studies the rules of the connection between nodes in social network, and compares and analyzes the prediction effect of several typical link prediction algorithms in social network. Aiming at the shortcomings of existing link prediction algorithms based on random walk, a neighbor modified algorithm for link prediction of random walk is proposed. The proposed algorithm solves the single neighbor selection situation which may occur in the original random walk algorithm. Considering the contribution of all neighbor nodes of the source node to the final similarity, the similarity between the original nodes is modified. The prediction algorithm of random walk link after neighbor correction can obtain good prediction effect, and the algorithm has lower time complexity and higher algorithm stability. The algorithm is applied to the recommendation of social network friends. On the other hand, from the point of view of social interpersonal network reconstruction, this paper analyzes and realizes the reconstruction of detected interpersonal network. Aiming at the shortcomings of the method of interpersonal network detection for Chinese text, the model which can reveal the relationship between the characters in the text is improved and obtained. Taking the text of "the three Kingdoms" as the research object, the relationship network between the characters in the three Kingdoms period is obtained by the abstract character model, and then the link prediction algorithm based on neighbor correction is used to predict the relationship. According to the predicted results, the detection network is reconstructed. It is found that the reconstructed network can reveal the relationship between the characters of the three Kingdoms more truthfully. The proposed method of random walk link prediction based on neighbor correction is helpful for people to understand the evolution mechanism of social network and the driving force of the relationship between people, and to realize the recommendation of friends in social network. The idea of network reconstruction is of great significance to the construction of real social network.
【学位授予单位】：东北大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP391.1;TP393.09

【相似文献】