摘要
在优化分类技术的研究中,文本特征化后通常具有高维性和不平衡性的特点,导致传统的分类算法准确率不高的问题。针对文本分类器的性能容易受到核函数和参数的影响的问题,为提高文本分类器的准确性。采用支持向量机(SVM)的理论在文本分类技术同时将根据优化的粒子群算法(PSO)引入SVM分类算法中进行优化文本分类器的参数,将分类器的准确率作为PSO算法适应度函数通过粒子移动操作找出最佳参数并用SVM算法进行分类。在文本数据集上的仿真结果表明,与传统的算法相比,经PSO算法优化后的SVM文本分类器的准确性更高,PSO算法是一种有效的优化方法,能广泛应用于文本分类问题。
Text characterization usually has the characteristics of high dimensional and unbalanced, which causes the probems that traditional classification algorithm accuracy is not high, the performance of text categorization is vulnerable to the influence of kernel function and parameters. In order to improve the accuracy of the text classifier, this article used the support vector machine (SVM) theory to study the text classification technology, and the theory of particle swarm optimization (PSO) algorithm, the classification algorithm was introduced to the SVM to optimize the parameters of the text classifie, we used the accuracy of the classifier as fitness functions, used particles move operation to find the best parameters, and used the SVM algorithm to classify the texts. Compared with the traditional algorithm, the new classifier has higher accuracy.
出处
《计算机仿真》
CSCD
北大核心
2013年第2期299-302,368,共5页
Computer Simulation