【摘要】 当前社交网络中普遍存在带有正面、负面属性的链接指向,构成了具有正负向关系的符号网络。由于传统的PageRank算法无法直接应用于这种新型网络环境,本文提出了一种计算符号网络中节点正、负影响力的改进算法。该算法同时考虑链接的正面和负面指向关系,通过分析节点被正向、负向访问的概率来确定节点在社交网络中的正、负影响力。小规模网络举例说明与Slashdot网站实际数据模拟分析的结果显示,改进算法符合符号网络中同时存在正负关系的特性,与PageRank算法在正负影响力的计算结果上存在差异。较之传统算法计算节点影响力时分离正负关系的做法,改进算法具备理论和实践的创新性。更多还原

【Abstract】 Today’s social networks include both positive and negative relations.Network with positive and negative relations is defined as signed network.Since traditional PageRank cannot be directly applied to signed networks,an improved algorithm is developed to calculate social actors’ positive and negative influence power.The algorithm improves PageRank by distinguishing the sign of social relationships and ranking social actors based on the probability of positive and negative visits.A small-scale network example and simulations based on the real Slashdot network suggest that compared to analyze positive and negative relations in separation,the improved algorithm accounts for the mixture of positive and negative interactions in signed network,and produces significantly different ranking results in contrast with traditional PageRank.The improved algorithm provides both theoretical insights and practical implications.更多还原