基于PSO-SVM算法的长微博贴图识别方法

微博由于字数的限制,当用户需要发较多内容时通常以附图的形式给出,识别包含文本内容贴图的长微博能够为微博研究提供更多有用的数据.在支持向量机（SVM）的基础上结合粒子群算法（PSO）提出了一种识别长微博贴图的PSO-SVM算法.该方法提取长微博贴图的颜色矩和灰度共生矩阵特征,然后利用PSO算法对SVM模型中的误差惩罚参数和核函数进行优化得到最佳分类模型,其最优参数将被用作长微博贴图和非长微博贴图进行分类.实验表明,与传统的基于网格搜索法优化的SVM算法相比,PSO-SVM算法对长微博贴图识别具有更高的准确率和召回率.

Identifying Images Representing Long Microblog Based on PSO-SVM Algorithm

Due to the length limitation of micro-blogs,users have to post images containing original text contents when they want to post long micro-blogs.If such images representing long micro-blog can be identified,it will provide more useful information for micro-blog research.An identification method based on support vector machine （SVM） and particle swarm optimization （PSO） is proposed in the paper.Firstly,the color moments and gray level concurrence matrix is extracted from image representing long micro-blog,then the error penalty parameter and kernel function of SVM are optimized by PSO algorithm to obtain the optimal classification model.The results show that,compared with traditional SVM based on grid searching method,the PSO-SVM algorithm has higher accuracy and recall rate of identifying images representing long microblog.