基于KK-SVM的垃圾标签检测方法研究-计算机软件与理论专业论文.docxVIP

本文将降维思想引入垃圾标签检测模型的数据约减阶段，提出了用 KPCA-SVM 检测垃圾标签的方法，这是本文的一个理论创新点。再将模 型实例化以后，通过核 K-means 聚类算法和降维算法的有机结合，生成 了效果更优的数据约减方法，最后形成了基于 KK-SVM 的垃圾标签检测 模型。将该模型首次实际运用到 Folksonomy 系统中，通过实验论证，基 于 KK-SVM 的垃圾标签检测模型比传统同类检测模型检测效果更好。这 是本文的一个应用创新点。 关键词：社会分类法 垃圾标签 数据降维 支持向量机 聚类 数据约减 模型 THE RESEARCH OF SOCIAL SPAM DETECTION METHOD BASED ON KK-SVM ABSTRACT Computer data almost abstracted from the objective world, because of the complexity of the objective world, data have extremely complex structure and high dimension, which no doubt will increase the processed difficulty of the computer data. Folksonomy is a new classification method developed through the custom tags, including users, described resource and identifier. Because the sexual characteristic of Folksonomy is freedom, it achieve the higher degree of resource sharing, but also lead to Folksonomy system often contains social spam. Folksonomy system dependent on user behavior too much, once part of the users appear abnormal behavior in the system, create social spam, it will be a serious threat to the operation of the whole system. In order to reduce the risk of system, it is necessary to detect the social spam. This paper builds social spam detection model through the theory of support vector machine (SVM), the principle is that using support vector machine (SVM) to test the user behavior and to make the classification, determining if a user put in social spam, than reduce social spam by limiting the illegal users behavior. Then inspired by kernel principal component analysis (kpca), introducing the ideas of data dimension reduction to reduct data sets, the raw data translate into low-dimensional data through dimension reduction, construct the method of social spam detection model based on KPCA-SVM. After that, using the kernel K-means clustering algorithm to optimize KPCA-SVM social spam detection model. Finally formed the social spam detection model based on KK-SVM. Among them, the previous K represent kernel K-means clustering a