基于VSM的海量医学资源特定信息优化聚类模型  

Optimal Clustering Model for Specific Information of Massive Medical Resources Based on VSM

作　　者：刘鹏[1] 宁鹏飞[1] LIU Peng;NING Peng-fei(Inner Mongolia Medical University,Huhehaote Inner Mongolia 010000,China)

机构地区：[1]内蒙古医科大学,内蒙古呼和浩特010000

出　　处：《计算机仿真》2021年第6期383-386,共4页Computer Simulation

摘　　要：目前医学资源信息聚类方法由于没有对特定信息进行去噪,使细节信息流失,无法保留有效信息,导致Jaccard系数与F1系数偏低,特定信息聚类效果较差的问题。为解决上述问题,提出基于VSM的海量医学资源特定信息优化聚类模型,采用有标记医学信息与无标记医学信息样本中所包含的信息,设置降维目标函数的参数值,通过建立降维矩阵实现医学资源特定信息降维处理;利用小波变换模极大值对医学资源特定信息去噪处理,在去噪过程中设定阈值,保留细节信息;在VSM的基础上建立医学资源特定信息优化聚类模型,实现海量医学资源特定信息的聚类。实验结果表明,所提方法Jaccard系数与F1系数较高,表明聚类结果与原有的类别系统更接近,即聚类效果的质量更好。Presently, due to the neglect of specific information de-noising, the detailed information of medical resource information clustering methods is seriously lost, and the effective information cannot be retained, resulting in low Jaccard and F1 coefficients and poor specific information clustering effect. Therefore, this paper put forward a VSM based optimal clustering model for specific information of massive medical resources. The information containing in the labeled and unlabeled medical information samples was used to set the parameter values of the dimension reduction objective function. The dimension reduction matrix was established to reduce the specific information of medical resources. The specific information of medical resources was de-noising via the maximum value of the wavelet transform model. During denoising, the threshold was set and the details were preserved. Based on VSM, the optimized clustering model of specific information of medical resources was established to achieve the clustering of specific information of massive medical resources. The results show that the method has high Jaccard and F1 coefficients, being close to the original classification system, implying an excellent clustering effect.

关 键 词：医学资源特定信息 聚类模型 信息预处理 特征提取 

分 类 号：TP391.12[自动化与计算机技术—计算机应用技术]