摘要
针对目前垃圾短信过滤效果有待提高的问题,提出一种新的短信特征提取方法。该方法采用了建立在深度学习理论基础上的最新成果和Word2vec工具。基于中文短信的内容和结构特点,利用该工具设计了一个短信向量化算法。该算法能有效地将每条短信与一个向量对应,在深度置信网络上利用该算法对垃圾短信进行分类实验。实验结果表明,推广性能比已有报道结果提高了约5%。
This paper proposes a new method of feature extraction of SMS for better spam message filtering.The method uses the latest results and tools of Word2 vec based on deep learning theory. With the content and structure characteristics of Chinese short messages in mind,an algorithm of Vectoring SMS is designed based on this tool.The algorithm can effectively match each text message with a vector. The classification＇s experiments on the spam messages are carried out using the proposed algorithm on the deep belief networks. The results show that the performance of the proposed algorithm is improved by 5% compared with the previously reported results.
出处
《电子科技》
2016年第4期49-52,共4页
Electronic Science and Technology
基金
国家自然科学基金资助项目(60443004)
校内科研基金资助项目(2014xcxtd05
2014xzky05)
关键词
深度置信网络
深度学习
短信
向量化
deep belief nets
deep learning
short messages
vectoring
