摘要
CDN的缓存策略对互联网应用的服务质量起着重要作用。针对传统的缓存策略通常仅考虑用户访问行为热度的情况,为了进一步优化缓存配置,提出通过挖掘用户评论中的文字信息来改善CDN的内容缓存替换策略。提出的方法在用户访问行为的基础上,融合用户基于文字评价的情感信息进行建模,并采用深度卷积神经网络对用户评论情感程度进行量化分析。实验结果表明,提出的CDN内容缓存替换方法在本地命中率方面较传统方法有很大程度的提高。
CDN caching strategy plays an important role in the quality of service of Internet applications. According to traditional caching strategies in which only users' access behavior is considered, the cache configuration is further optimized that the text information in users' comments is mined to improve the CDN content cache replacement strategy. Based on the users' access behavior, the proposed method integrates the sentiment information of users' comments to establish model. The deep convolutional neural network is used to quantitatively analyze the sentiment of users' comments. Experimental results show that the proposed CDN content cache replacement method is greatly improved in the hit rate compared with the traditional method.
作者
陈步华
陈戈
梁洁
CHEN Buhua;CHEN Ge;LIANG Jie(Gnangzhou Research Institute of China Telecom Co., Ltd., Guangzhou 510630, China)
出处
《移动通信》
2018年第5期80-84,共5页
Mobile Communications
基金
国家高技术研究发展计划("863"计划)基金资助项目"动态媒体业务支撑平台应用与示范"(2015AA015803)
关键词
评论情感分析
内容分发网络
卷积神经网络
缓存替换
sentiment analysis content distribution network
convolutional neural network
caching replacement