计算机与数字工程2017，Vol.45Issue(10)：1990-1995,6.DOI:10.3969/j.issn.1672-9722.2017.10.021
基于文本上下文和网络信息的链接预测方法
Link Prediction Method Based on Context and Network Information
任奕豪 1张琨 1赵静 1冯新淇1
作者信息
- 1. 南京理工大学计算机科学与工程学院 南京210094
- 折叠
摘要
Abstract
In regard to link prediction problem,traditional prediction models usually only consider the link information of the nodes from the network.However,the text widely existing in the social networks can be used to improve the performance of link pre-diction,and using text for link prediction is getting attention increasingly.Combining text and links,proposes a link prediction mod-el based on hierarchical latent dirichlet allocation topic model.First,the model trains text data by hierarchical latent dirichlet alloca-tion model,then it extracts text similar features from the convergent topic tree,finally the model trains the feature data to obtain a two-class classifier by support vector machine model,this classifier can be used to predict the link between nodes.The experimen-tal results demonstrate that,comparing to pre-existing similar models,the model proposed improves the accuracy of predicting the links among the documents in text network.关键词
链接预测/层次隐狄利克雷分布/主题树/文本相似特征/支持向量机Key words
link prediction/hierarchical latent dirichlet allocation/topic tree/text similar feature/support vector machine Class Number TP301分类
信息技术与安全科学引用本文复制引用
任奕豪,张琨,赵静,冯新淇..基于文本上下文和网络信息的链接预测方法[J].计算机与数字工程,2017,45(10):1990-1995,6.
