| 常芳瑞,王增光,贾萧丽,王子琪,袁学旺.基于深度学习的谣言检测研究[J].情报工程,2026,(2):015-027 |
| 基于深度学习的谣言检测研究 |
| Research on Rumors Detection Based on Deep Learning |
| |
| DOI： |
| 中文关键词: 谣言检测；深度学习；谣言数据集；神经网络 |
| 英文关键词: Rumor Detection; Deep Learning; Rumor Data Sets; Neural Networks |
| 基金项目: |
| 作者 | 单位 | | 常芳瑞 | 山东省互联网舆情中心 济南 250000 | | 王增光 | 潍坊市互联网舆情中心 潍坊 261000 | | 贾萧丽 | 潍坊市互联网舆情中心 潍坊 261000 | | 王子琪 | 山东省互联网舆情中心 济南 250000 | | 袁学旺 | 潍坊市互联网舆情中心 潍坊 261000 |
|
| 摘要点击次数: 33 |
| 全文下载次数: 23 |
| 中文摘要: |
| [目的/意义]网络时代下社交媒体使用更加广泛，互联网谣言随之大量涌现，给社会和公众带来诸多不利影响。如何准确、及时地发现谣言，学术界进行了大量研究和探索。传统谣言检测方法提取数据特征较为单一，而深度学习预测方法能够处理更复杂的特征结构，成为谣言检测的研究热点。[方法/过程]文章对基于深度学习模型的谣言检测方法进行综述，详细介绍了深度学习架构循环神经网络、卷积神经网络、图卷积神经网络、注意力机制以及混合神经网络。其次，总结了现有文献中常用的国内外公共数据集。再次，根据前人的模型实验对谣言检测模型的性能和优劣、改进方向进行了分析。最后，探讨了深度学习谣言检测领域未来的研究方向。[局限] 未深入对各类模型的超参数和使用场景分析进行研究，后续需要进一步总结。[结果/结论]对近年来各类深度模型在谣言检测方面的应用进行了系统梳理，帮助研究者快速了解研究范围和重点，为后续选择合适的模型提供一定参考。 |
| 英文摘要: |
| [Objective/Significance] In the Internet era, social media is more widely used, and rumors on the Internet have emerged in large quantities, bringing many adverse effects to society and the public. The academic community has conducted a lot of research and exploration on how to accurately and promptly discover rumors. Traditional rumor detection methods extract data characteristics are relatively single, while deep learning prediction methods can handle more complex feature structures,becoming a hot topic in rumor detection. [Methods/Processes] The article reviews the rumor detection method based on deep learning models, and introduces in detail the deep learning architecture recurrent neural network, convolutional neural network,graph convolutional neural network, attention mechanism and hybrid neural network. Secondly, the commonly used domestic and foreign public data sets in the existing literature are summarized. Again, based on the model experiments of previous generations, the performance, advantages and disadvantages of the rumor detection model and the direction of improvement were analyzed. Finally, the future research directions in the field of deep learning rumors detection are discussed. [Limitations] The hyperparameters and usage scenario analysis of various models have not been further discussed, and further summary is needed in the future. [Results/Conclusions] The article systematically sorted out the applications of various in-depth models in rumors detection in recent years, helping researchers quickly understand the scope and focus of the research, and providing a certain reference for subsequent selection of suitable models. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
