基于结构平衡理论及LP算法的符号网络预测

符号网络分析逐渐成为一个越来越重要的研究主题,其中最为重要的是网络中的符号推断问题.了解到在社会网络中局部路径指标(LP)表现良好.用AUC评价指标进行了实验验证,LP指标同样适用于符号网络,并且给出它的平衡理论解释.实际数据分析的结果显示,较之传统计算符号网络链路预测算法,该方法更加简单,并且能得到较好的预测效果.

Signed network prediction based on structural balance theory and LP algorithm

In recent years,signed network analysis is becoming a more and more important research topic. Link prediction is a key one in the research of signed network. It is known that the classic LP measure is used widely in unsigned link prediction. This paper gives an interpretation of LP index with the balance theory when applied to signed networks. It gives an experimental verification,including comparisons with some local approaches with AUC evaluation index. The experimental results indicate that LP index is simpler and can get better prediction results than the traditional algorithm of signed networks.