融合多维属性相似度的云制造服务谱聚类算法  

Cloud manufacturing service spectrum clustering algorithm based on multi-dimensional attribute similarity

作　　者：黄雯 胡强[1] 任志考[1] HUANG Wen;HU Qiang;REN Zhi-kao(College of Information Science and Technology,Qingdao University of Science and Technology,Qingdao 266061,China)

机构地区：[1]青岛科技大学信息科学技术学院,山东青岛266061

出　　处：《计算机工程与设计》2023年第10期2988-2996,共9页Computer Engineering and Design

基　　金：国家自然科学基金项目(61973180);山东省自然科学基金项目(ZR2019MF033);山东省重点研发计划软科学基金项目(2021RKY02037);云南省教育厅科学研究基金项目(2022J0635)。

摘　　要：为提高云制造服务的聚类质量,提出一种融合多维属性相似度的云制造服务谱聚类算法。将云制造服务的文本型属性值向量化和数值型属性值归一化,分别构建文本型属性相似度矩阵和数值型属性相似度矩阵。设计一种属性相似度融合函数,实现多维度云制造服务属性相似度的融合,建立云制造服务相似度矩阵。引入相似矩阵的本征间隙确定聚类个数,实现高效、精准的云制造服务谱聚类。实验结果表明,该方法较其它聚类方法显著提高了云制造服务的聚类质量。To improve the quality of cloud manufacturing services,a spectrum clustering algorithm integrating multi-dimensional attribute similarity was proposed.The text attribute values of cloud manufacturing services were vectorized to construct the text attribute similarity matrix.The numerical attribute values were normalized to build numerical attribute similarity matrix.An attribute similarity fusion function was designed to achieve the fusion of multi-dimensional attribute similarities.The cloud manufacturing service similarity matrix was built on the fused attribute similarities.The eigengap of the service similarity matrix was introduced to determine the number of service clusters.The efficiency and accuracy of cloud manufacturing service spectrum clustering were improved.Compared with other spectral clustering methods,the experiment results show that the proposed method significantly enhances the service clustering quality.

关 键 词：云制造服务 谱聚类 本征间隙 相似度矩阵 融合函数 文本向量 归一化 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]