基于流式计算的垃圾短信治理关键技术研究  

Research on Key Technologies for Spam SMS Management Based on Streaming Computing

作　　者：王九九 狄秋燕 马永亮[1] Wang Jiujiu;Di Qiuyan;Ma Yongliang(China United Network Communications Group Co.,Ltd.,Beijing 100033,China)

机构地区：[1]中国联合网络通信集团有限公司,北京100033

出　　处：《邮电设计技术》2024年第5期56-61,共6页Designing Techniques of Posts and Telecommunications

摘　　要：某运营商在现网垃圾短信治理中,常采用关键字+规则的方法,难以在拦截成功率和误拦正常短信之间找到平衡。基于文本语义分析识别垃圾短信,则需要解决大数据挖掘算法、海量数据处理、响应时效等问题,因此在大业务量的集约化平台上应用并不广泛。通过算法研究、开发原型系统等工作,探索基于流式计算的垃圾短信治理技术方案,研发了一套基于Storm+Mahout架构的垃圾短信识别原型系统,完成了性能和准确率测试,取得了较好的效果。A certain operator often adopts the method of keyword+rule in the management of spam messages on the current network,which makes it difficult to strike a balance between the success rate of intercepting spam messages and the error rate of normal messages.Based on text semantic analysis to identify spam messages,it is necessary to solve problems such as big data mining algorithms,massive data processing,and response time.Therefore,it is less applied on intensive platforms with large business volumes.It explores a spam message management technology solution based on streaming computing through algorithm research and prototype system development.A spam message recognition prototype system based on Storm+Mahout architecture has been developed,and the performance and accuracy tests have been completed,achieving good results.

关 键 词：垃圾短信治理 自然语言处理 大数据 流式计算 

分 类 号：TN919[电子电信—通信与信息系统]