基于元学习的萌动期网络舆情预测模型
Public Opinion Prediction Model Based on Meta-Learning
DOI: 10.12677/mos.2024.133201, PDF,    科研立项经费支持
作者: 刘吉金, 曹春萍：上海理工大学光电信息与计算机技术学院，上海
关键词: 网络舆情；舆情预测；深度学习；元学习；双向长短期记忆网络；Internet Public Opinion； Public Opinion Prediction； Deep Learning； Meta-Learning； BiLSTM
摘要: 现有的时间序列预测模型在网络舆情处于萌动期时，因训练所用数据大多为历史相似事件的网络舆情数据，存在严重的过拟合问题。针对上述问题，本文提出一种基于元学习的萌动期网络舆情预测模型Reptile-BiLSTM，基于SIR模型构建舆情传播网络，根据舆情特点优化OLEI算法并计算舆论的影响力特征，在历史相似事件的舆情数据上对BiLSTM模型进行预训练，使用Reptile算法在少量目标事件的舆情数据上对预训练模型进行微调，提高模型在萌动期网络舆情发展预测任务上的表现。经实验验证，Reptile-BiLSTM在evs、r2指标上较BiLSTM分别上升6%、13%，表明本文提出的模型能够在网络舆情处于萌动期时较准确地预测其发展趋势，能够尽早地为相关工作人员引导舆情发展提供有效的决策支持。
Abstract: When the online public opinion is in the germination stage, the existing time series prediction model has a serious overfitting problem because most of the training data are the online public opinion data of similar historical events. To solve the above problems, this paper proposes a meta-learning based online public opinion prediction model Reptile-BiLSTM, constructs a public opinion communication network based on the SIR Model, optimizes the OLEI algorithm according to the characteristics of public opinion, calculates the influence characteristics of public opinion, and pre-trains the BiLSTM model on the public opinion data of similar historical events. The Reptile algorithm is used to fine-tune the pre-trained model on the public opinion data of a small number of target events to improve the model’s performance in predicting the development of online public opinion during the budding period. The experimental results show that the evs and r2 indexes of Reptile-BiLSTM are 6% and 13% higher than those of BiLSTM respectively, indicating that the model proposed in this paper can accurately predict the development trend of network public opinion when it is in the sprouting stage, and provide effective decision support for relevant staff to guide the development of public opinion as soon as possible.
文章引用：刘吉金, 曹春萍. 基于元学习的萌动期网络舆情预测模型[J]. 建模与仿真, 2024, 13(3): 2189-2202. https://doi.org/10.12677/mos.2024.133201

参考文献

[1] 谢科范, 赵湜, 陈刚, 等. 网络舆情突发事件的生命周期原理及集群决策研究[J]. 武汉理工大学学报(社会科学版), 2010, 23(4): 482-486.
[2] Siami-Namini, S., Tavakoli, N. and Namin, A.S. (2019) The Performance of LSTM and BiLSTM in Forecasting Time Series. 2019 IEEE International Conference on Big Data (Big Data), Los Angeles, 9-12 December 2019, 3285-3292. [Google Scholar] [CrossRef] 
[3] Nichol, A., Achiam, J. and Schulman, J. (2018) On First-Order Meta-Learning Algorithms. arXiv: 1803.02999.
[4] Kermack, W.O. and Mckendrick, A.G. (1991) Contributions to the Mathematical Theory of Epidemics—III. Further Studies of the Problem of Endemicity. Bulletin of Mathematical Biology, 53, 89-118. [Google Scholar] [CrossRef] 
[5] 王晰巍, 毕樱瑛, 李玥琪. 社交网络中意见领袖节点影响力指数模型及实证研究——以自然灾害“7∙20”河南暴雨为例[J]. 图书情报工作, 2022, 66(16): 24-35. [Google Scholar] [CrossRef] 
[6] Hochreiter, S. and Schmidhuber, J. (1997) Long Short-Term Memory. Neural Computation, 9, 1735-1780. [Google Scholar] [CrossRef] [PubMed]
[7] Graves, A. and Schmidhuber, J. (2005) Framewise Phoneme Classification with Bidirectional LSTM Networks. Proceedings. 2005 IEEE International Joint Conference on Neural Networks, Montreal, 31 July-4 August 2005, 2047-2052.
[8] Chung, J., Gulcehre, C., Cho, K.H. and Bengio, Y. (2014) Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling. arXiv: 1412.3555.
[9] Bahdanau, D., Cho, K. and Bengio, Y. (2014) Neural Machine Translation by Jointly Learning to Align and Translate. arXiv: 1409.0473.
[10] 白静, 李霏, 姬东鸿. 基于注意力的BiLSTM-CNN中文微博立场检测模型[J]. 计算机应用与软件, 2018, 35(3): 266-274.
[11] 黄炜, 黄建桥, 李岳峰. 基于BiLSTM-CRF的涉恐信息实体识别模型研究[J]. 情报杂志, 2019, 38(12): 149-156.
[12] Xu, G., Meng, Y., Qiu, X., Yu, Z. and Wu, X. (2019) Sentiment Analysis of Comment Texts Based on BiLSTM. IEEE Access, 7, 51522-51532. [Google Scholar] [CrossRef] 
[13] Cai, R., et al. (2020) Sentiment Analysis about Investors and Consumers in Energy Market Based on BERT-BiLSTM. IEEE Access, 8, 171408-171415. [Google Scholar] [CrossRef] 
[14] 曾子明, 孙晶晶. 基于用户注意力的突发公共卫生事件舆情情感演化研究——以新冠肺炎疫情为例[J]. 情报科学, 2021, 39(9): 11-17.
[15] Breitfeller, L., Naik, A. and Rose, C. (2021) STAGE: Tool for Automated Extraction of Semantic Time Cues to Enrich Neural Temporal Ordering Models. arXiv: 2105.07314.
[16] Nandwani, P. and Verma, R. (2021) A Review on Sentiment Analysis and Emotion Detection from Text. Social Network Analysis and Mining, 11, Article No. 81. [Google Scholar] [CrossRef] [PubMed]
[17] 陈登建, 杜飞霞, 杨秀璋, 夏换. 基于LDA-CNN-BiLSTM与四象限法则的突发公共事件舆情主题演化研究[J]. 情报探索, 2022(10): 45-53.
[18] 刘忠宝, 秦权, 赵文娟. 微博环境下新冠肺炎疫情事件对网民情绪的影响分析[J]. 情报杂志, 2021, 40(2): 138-145.
[19] Mohapatra, A., Thota, N. and Prakasam, P. (2022) Fake News Detection and Classification Using Hybrid BiLSTM and Self-Attention Model. Multimedia Tools and Applications, 81, 18503-18519. [Google Scholar] [CrossRef] 
[20] 曾莉, 杨添宝, 周慧. 基于LDA与注意力机制BiLSTM的微博舆情分析模型[J]. 南京理工大学学报, 2022, 46(6): 742-748.
[21] 李德宇. 新媒体背景下的网络舆情传播研究综述[J]. 社会科学与计算研究, 2022, 2(1): 13-22.
[22] 陈珂, 刘莉. 话题分析和情感分析视角下的高校网络舆情衍化特征研究[J]. 全媒体探索, 2022(10): 64-69.
[23] Wang, L., Zhang, X., Su, H. and Zhu, J. (2023) A Comprehensive Survey of Continual Learning: Theory, Method and Application. arXiv: 2302.00487.

为你推荐