基于语音话单分析的骚扰电话识别技术研究  

Harassment Call Recognition Technology Research Based on Voice Bill Analysis

作　　者：李正 陶冶[2] Li Zheng;Tao Ye(The 6th Research Institute of China Electronics Corporation,Beijing 100083,China;China Unicom Research Institute,Beijing 100176,China)

机构地区：[1]中国电子信息产业集团有限公司第六研究所,北京100083 [2]中国联通研究院,北京100176

出　　处：《信息通信技术》2021年第6期18-25,共8页Information and communications Technologies

摘　　要：文章基于某运营商现网话单数据,采用机器学习方法,提出一种基于语音话单分析的骚扰电话识别模型,并对模型进行了验证。验证结果表明,该模型能够有效识别骚扰电话,与互联网企业相比具有识别范围广、准度高的特点。同时,文章与运营商的现网业务相结合,总结出四种骚扰电话提醒方案,并对每个方案的优缺点进行说明,以期对运营商开展骚扰电话提醒业务具有一定的借鉴和指导意义。In this paper, a harassment call recognition model is designed and proposed based on machine learning algorithms and call bill data of a telecom operator’s current network. Then, the model is validated, and the result shows that the model can effectively identify harassing calls, and it has wider identification range and higher accuracy comparing to big technology companies. Meanwhile, this paper combines the telecom operators’ businesses to propose four harassment call reminder schemes. Next, the advantages and drawbacks of each scheme are explained in order to have reference and guiding significance for the telecom operators’ harassment call reminder services.

关 键 词：大数据 信息安全 机器学习 语音话单 骚扰电话 

分 类 号：D922.16[政治法律—宪法学与行政法学] D924.3[政治法律—法学] TP181[自动化与计算机技术—控制理论与控制工程]