摘 要:网络舆情发展迅速,有效地实现话题的热度预测对网络舆情监管和正确引导具有重要意义。为了实现对话题热度的预测,并提高预测的精度,本文提出基于小波神经网络的话题热度预测模型。首先经过预处理分析发现热点话题,然后选取具有热度表征能力的指标量化话题热度,构建出话题热度指标的时间序列模型,最后提出基于小波神经网络模型的话题热度预测方法,预测出下一时间段的话题热度值。实验结果显示,与BP 神经网络预测模型相比,本文提出的小波神经网络预测模型无论从预测曲线的拟合度还是绝对误差标准差的数值上都具有更高的预测精度及稳定性。
关键词:网络舆情;热点话题;热度预测;小波神经网络
中图分类号:TP391 文献标识码:A 文章编号:2096-4706(2018)05-0074-05
Research on Topic Heat Prediction Model Based on Wavelet Neural Network
TAN Peng,LUO Shunlian,SUN Xiaosong,WANG Hui,LIANG Xiaohan
(Civil Aviation University of China,Tianjin 300300,Chima)
Abstract:With the rapid development of network public opinion,realizing the forecast of hot topic has great importance for supervision and correct guidance of network public opinion. In order to realize the prediction of the hot topic and improve the accuracy of prediction. This paper proposes a prediction model based on wavelet neural network. Firstly,analysis the text that has been pretreated to find the hot topic,and then we select some typical characterizations of heat topic index to construct the time series of heat topic value. Finally,a topic heat prediction method based on wavelet neural network model is proposed to predict the topic heat value of the next period. The experimental results show that compared with BP neural network the prediction model proposed in this paper has higher prediction accuracy and stability whether in curve fitting or the numerical absolute error of standard deviation.
Keywords:internet public opinion;hot topic;heat prediction;wavelet neural network
参考文献:
[1] 刘洋. 新媒体背景下涉警网络舆情的应对与引导 [J]. 法制与社会,2017(5):201-202.
[2] 张雪婷. 企业面对网络舆情的回应博弈策略研究 [J]. 金融经济,2017(2):125-126.
[3] 曹玖新,吴江林,石伟,等. 新浪微博网信息传播分析与预测 [J]. 计算机学报,2014,37(4):779-790.
[4] 侯凯. 微博中话题的传播模型及热点预测 [D]. 北京:北京邮电大学,2015.
[5] 张博,李竹君. 微博信息传播效果研究综述 [J]. 现代情报,2017,37(1):165-171.
[6] 刘勘,李晶,刘萍. 基于马尔可夫链的舆情热度趋势分析 [J]. 计算机工程与应用,2011,47(36):170-173.
[7] 邓青,马晔风,刘艺,等. 基于BP 神经网络的微博转发量的预测 [J]. 清华大学学报(自然科学版),2015,55(12):1342-1347.
[8] 聂恩伦,陈黎,王亚强,等. 基于K 近邻的新话题热度预测算法 [J]. 计算机科学,2012,39(S1):257-260.
[9] 曾振东. 基于灰色支持向量机的网络舆情预测模型 [J]. 计算机应用与软件,2014,31(2):300-302+311.
[10] 陈羽中,方明月,郭文忠,等. 基于小波变换与差分自回归移动平均模型的微博话题热度预测 [J]. 模式识别与人工智能,2015,28(7):586-594.
[11] 闫辉辉,朱智慧,刘伦铭,等. 基于小波神经网络时间序列模型预测血药浓度的研究 [J]. 中国现代应用药学,2016,33(11):1417-1422.
[12] Jin Ran,Li Shengcai. Comparison of occurrence periodsof wheat aphids based on artifical neural network and wavelet neural network prediction systems [J]. Journal of Plant Protection,2016,43(3):353-361.
作者简介:谭鹏(1996-),女,汉族,四川人,研究方向:舆情分析、软件。