Networked data classification in social media based on manifold learning
基于流形学习的社会化媒体网络数据分类

Social media provided massive, large-scale heterogeneous networked data. Classification in networked data is a new problem that needed to be solved. Based on latent social dimension model, this paper proposed using Laplacian eigenmaps from manifold learning to extract social dimensions . Experiments show that it is superior to original modularity maximization social dimension model in performance metrics like exact match ratio, micro average and macro average. The algorithm can capture implicit user relations better and analysis Web user behavior better.