基于机器学习的改进型物联网服务信任模型  

Improved trust model of internet of things services based on machine learning

作　　者：何超勋 彭伟锋 李燕飞 陈海彪 HE Chao-xun;PENG Wei-feng;LI Yan-fei;CHEN Hai-biao(Information Center,Guangdong Power Grid Co.,Ltd.Shanwei Power Supply Bureau,Shanwei 516600,China;School of Computer Science and Engineering,South China University of Technology,Tianhe 510641,China)

机构地区：[1]广东电网有限责任公司汕尾供电局信息中心,广东汕尾516600 [2]华南理工大学计算机科学与工程学院,广州天河510641

出　　处：《计算机工程与设计》2022年第5期1335-1343,共9页Computer Engineering and Design

基　　金：国家自然科学基金项目(61771203、61803161)。

摘　　要：信任机制可以解决物联网信息中的威胁和隐患等问题,但由于影响因素众多,网络中的信任制度难以建立。因此,需要一个智能信任计算模型,能够为潜在的参与者生成准确和直观的信任值。在此提出一个可量化的信任评估模型,实现对个体信任属性进行数值计算。提出一种基于机器学习原理的信任特征分类算法,将提取的信任特征进行组合,得到最终的信任值,用于决策。实验结果表明,该模型是有效的,相对于其它方法,该方法性能更加优越。The trust mechanism can solve the threats and hidden dangers in the information of the internet of things,but due to many influencing factors,it is difficult to establish a trust system in the Internet.Hence,an intelligent trust calculation model is needed that can generate accurate and intuitive trust values for potential participants.Therefore,a quantifiable trust evaluation model was proposed to realize the numerical calculation of individual trust attributes.A trust feature classification algorithm based on machine learning principles was proposed.The extracted trust features were combined to obtain the final trust value for decision-making.Experimental results show that the model is effective,and compared with other methods,the performance of this method is superior.

关 键 词：信任模型 信任计算 特征提取 聚类 模型分类 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]