核K均值聚类改进神经网络垃圾标签检测模型

Model of Detecting Tag Spam Based on Kernel K-Means and Improved Neural Network

为了提高垃圾标签检测精度,提出一种核K均值聚类和改进神经网络相融合的垃圾标签检测模型。首先核K均值聚类算法提抽取垃圾标签的特征向量,然后将特征向量集输入到BP神经网络进行训练,并采用混沌粒子群算法对BP神经网络的参数进行优化,最后建立垃圾标签检测模型,并通过仿真实验对模型性能测试。结果表明,该垃圾标签检测算法模型不仅提高了垃圾标签识别率,训练时间大幅度减少,垃圾标签检测效率得到提高,可以较好满足垃圾标签实时、在线检测要求。

In order to improve the tag spam detection accuracy, this paper presents a tag spam detection model based on kernel K-means clustering and improved neural network. Firstly, the feature vector is extracted by kernel K-means clus-tering algorithm, and then the feature vector set is input to the BP neural network for training, and the parameters of the BP neural network is optimized by chaos particle swarm optimization algorithm, finally, the spam detection model is es-tablished, and the simulation experiment is carried out to test performance of the model. The results show that the pro-posed mode not only improves the tag spam recognition rate, and the training time is greatly reduced, spam detection effi-ciency is improved, it can better meet the requirements of real-time, on-line tag spam detecting.