基于磷虾群免疫神经网络的微博僵尸粉检测  

DETECTING MICROBLOGGING ZOMBIE FANS BASED ON KRILL HERD IMMUNE NEURAL NETWORK

作　　者：岳虹 张智 杨科 

机构地区：[1]61070部队,福建福州350001

出　　处：《计算机应用与软件》2015年第12期145-149,共5页Computer Applications and Software

摘　　要：为了更加准确地对微博僵尸粉进行甄别,提出基于磷虾群免疫神经网络的检测算法。首先,从静态与动态两个方面,分析并选取微博僵尸粉区别于普通用户的特征;其次,将磷虾群优化思想以及人工免疫的变异操作引入到网络连接权值和阈值的优化过程中,提高网络训练的收敛速度和泛化能力。最后,利用新浪微博数据,依靠训练后的神经网络对僵尸粉进行检测。实验结果表明,新算法具有更高的准确率和召回率,能够有效地检测出微博僵尸粉。In order to discriminate microblogging zombie fans more accurately,we proposed a novel detection algorithm which is based on krill herd immune neural network. First,we analysed and selected the features of microblogging zombie fans differing from ordinary users from static and dynamic aspects. Then,we introduced both the krill herd optimisation idea and the artificial immune mutation operation into the optimisation process of network connection weights and thresholds,so the convergence speed and generalisation ability of the neural network model were improved. Finally,based on Sina microblogging data we detected the zombie fans by trained neural network. Experimental results indicated that the novel algorithm had higher accuracy rate and recall rate,and it could detect the zombie fans effectively.

关 键 词：微博僵尸粉 检测 磷虾群 人工免疫 神经网络 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]