摘要
针对传统LDA主题模型忽视节点重要性的问题,提出一种新的社会网络话题发现算法iMLDA(Importance-Latent Dirichlet Allocation)。该算法利用传统的LDA主题模型,与基于Pagerank的节点重要性算法相融合,充分挖掘社会网络中蕴含的结构信息,提高传统LDA算法模型中携带的信息量,进而提高话题发现的准确率。实验结果证明,该算法取得了较好的实验结果。
For the problem that structure information of social networks may be ignored by the traditional LDA topic model, a novel topic discovery algorithm iMLDA（Importance-Latent Dirichlet Allocation） is proposed based on modified LDA model. The novel algorithm uses LDA model incorporating Pagerank algorithm,deeply mines the structure of social network ,in- creases the amount of information carried in traditional LDA model, and thus enhances the accuracy of topic discovery.The experiments show that the proposed model has a better effect than the traditional LDA model.
出处
《情报科学》
CSSCI
北大核心
2016年第9期115-118,133,共5页
Information Science
基金
国家自然科学基金项目(61502281
71403151)
博士点基金联合资助课题(20133718120011)
