基于多维相似度属性的社会网络链接预测算法研究

Research on Social Network Link Prediction Algorithm Based on Multidimensional Similarity Attributes

摘要: 链接预测是寻找社会网络中隐藏的和未来可能出现的链接，它对于分析社会网络具有重要意义。本文在对现有社会网络链接预测研究的基础上，分析了社会网络链接预测算法中的多维相似度属性，并把链接预测问题转换为分类问题，尝试使用机器学习的方法解决社会网络链接预测问题，最终通过实验得到验证，相似度属性特征对链接预测具有较高影响力，链接预测问题可以转化为分类问题通过机器学习算法得到解决。

Abstract: Link prediction refers to searching for hidden links or predicting possible future links in social networks. It is important for analyzing social networks. This paper analyzes the current methods for social network link prediction, compares multidimensional similarity attributes, takes the link prediction problem as a classification problem, and realizes links prediction based on machine learning. The final experiment results verify that similarity attributes are effective for link prediction problem, and link prediction problem can be solved as a classification problem by machine learning algorithms.