隨著網路使用普及，資訊可信度的重要性逐漸浮現。特別是在2016年美國總統大選期間，當時的候選人川普指責美國CNN、紐約時報等主流媒體報導假新聞，引起全球對於新聞可信度的討論與關注。這事件引發人們對於新聞的可信度，以及如何發現假新聞等問題的關注。

然而假新聞的問題不僅在於難以區分，也對社會造成嚴重危害。假新聞存在的情況下，人們往往會因為信任錯誤的消息而做出錯誤的判斷，進而對社會產生不良影響，突顯了訓練一個穩定的模型來區分假新聞非常重要。然而傳統深度學習方法能夠將不同視角映射到一個共同的特徵空間中，獲取視角間的交互資訊，但視角間也會競爭資源，而導致某些視角在完整特徵擷取前被忽略。

相較之下，本研究確保了多種視角在特徵提取方面具有足夠的資源，同時最大限度地降低了因特徵視角訓練速度不同而引起的欠擬合或過擬合現象。研究結果表明，使用多視角學習方法與模型複雜度分析，在LIAR假新聞資料集的準確度達到了當前的最先進水平 67.83%。

然而在2022年底 ChatGPT 問世後，其生成逼真的文本並模擬人類對話的能力，甚至能夠糾正使用者的不當提問，大幅降低了產生更多元、更多面向文本的成本。同時本研究發現，透過 ChatGPT 中立化重寫的文本，能使假新聞辨識模型，更加容易將中立化重寫的新聞判定為真實訊息。

為了充分發揮 ChatGPT 生成文本的優勢，本研究通過 ChatGPT 中立化假新聞資料集並結合重要性機制和加權平均嵌入方法，在二分類平衡 LIAR 假新聞資料集中，假新聞的召回率提高了5.26%，整體準確度提高了3.17%，提昇了假新聞偵測器在不同情緒強度下的穩健性。

With the widespread use of the internet, the importance of information credibility has become apparent. Particularly during the 2016 US presidential election, then-candidate Donald Trump accused mainstream media outlets like CNN and The New York Times of reporting fake news, sparking a global discussion and concern about the authenticity of news. This event raised awareness about the authenticity of news and how to detect fake news.

However, the problem of fake news extends beyond the difficulty of distinguishing it and poses serious harm to society. In the presence of fake news, people often make erroneous judgments based on false information, leading to adverse societal effects. This highlights the importance of training a robust model to distinguish fake news. However, traditional deep learning methods map different views to a common feature space, capturing the interaction information between views but also competing for resources among views, potentially neglecting certain views before complete feature extraction.

In contrast, this research ensures that multiple views have sufficient resources for feature extraction while minimizing underfitting or overfitting caused by different speeds of feature view training. The results show that using multi-view learning methods and model complexity analysis achieves state-of-the-art accuracy of 67.83\% on the LIAR fake news dataset.

With ChatGPT's debut in late 2022, its lifelike text generation and human-like conversation simulation, along with the ability to correct user queries, have markedly reduced the cost of creating diverse content. This research also noted that employing ChatGPT's neutralized rewrites makes fake news detectors more likely to classify them as genuine messages.

To harness ChatGPT's strengths, this study neutralized fake news data using ChatGPT and combined importance mechanisms and weighted average embedding. This improved the balanced binary classification on the LIAR fake news dataset with a 5.26% rise in fake news recall and a 3.17% boost in overall accuracy, enhancing the detector's robustness across emotional intensities.

[1] D. M. Lazer, M. A. Baum, Y. Benkler, A. J. Berinsky, K. M. Greenhill, F. Menczer, M. J. Metzger, B. Nyhan, G. Pennycook, D. Rothschild, et al., “The science of fake news,” Science, vol. 359, no. 6380, pp. 1094–1096, 2018.
[2] J. Devlin, M.-W. Chang, K. Lee, and K. Toutanova, “BERT: Pre-training of deep bidirectional transformers for language understanding,” in Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers), (Minneapolis, Minnesota), pp. 4171–4186, Association for Computational Linguis tics, June 2019.
[3] W. Y. Wang, ““liar, liar pants on fire”: A new benchmark dataset for fake news detection,” in Proceedings of the 55th Annual Meeting of the Association for Com putational Linguistics (Volume 2: Short Papers), (Vancouver, Canada), pp. 422–426, Association for Computational Linguistics, July 2017.
[4] B. Upadhayay and V. Behzadan, “Sentimental liar: Extended corpus and deep learn ing models for fake claim classification,” in 2020 IEEE International Conference on Intelligence and Security Informatics (ISI), pp. 1–6, IEEE, 2020.
[5] S. Hakak, M. Alazab, S. Khan, T. R. Gadekallu, P. K. R. Maddikunta, and W. Z. Khan, “An ensemble machine learning approach through effective feature extraction to classify fake news,” Future Generation Computer Systems, vol. 117, pp. 47–58, 2021.
[6] J. Y. Khan, M. T. I. Khondaker, S. Afroz, G. Uddin, and A. Iqbal, “A benchmark study of machine learning models for online fake news detection,” Machine Learning with Applications, vol. 4, p. 100032, 2021.
[7] V. Jain, R. K. Kaliyar, A. Goswami, P. Narang, and Y. Sharma, “Aenet: an attention enabled neural architecture for fake news detection using contextual features,” Neu ral Computing and Applications, vol. 34, p. 771–782, Aug 2021.
[8] M. Samadi, M. Mousavian, and S. Momtazi, “Deep contextualized text representa tion and learning for fake news detection,” Information Processing & Management, vol. 58, no. 6, p. 102723, 2021.
[9] M. H. Goldani, R. Safabakhsh, and S. Momtazi, “Convolutional neural network with margin loss for fake news detection,” Information Processing & Management, vol. 58, no. 1, p. 102418, 2021.
[10] O. Ajao, D. Bhowmik, and S. Zargari, “Sentiment aware fake news detection on online social networks,” in ICASSP 2019 - 2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pp. 2507–2511, 2019.
[11] R. Kumari, N. Ashok, T. Ghosal, and A. Ekbal, “What the fake? probing misin formation detection standing on the shoulder of novelty and emotion,” Information Processing & Management, vol. 59, no. 1, p. 102740, 2022.
[12] X. Yan, S. Hu, Y. Mao, Y. Ye, and H. Yu, “Deep multi-view learning methods: A review,” Neurocomputing, vol. 448, pp. 106–129, 2021.
[13] Y. Li, M. Yang, and Z. Zhang, “A survey of multi-view representation learning,” IEEE Transactions on Knowledge and Data Engineering, vol. 31, no. 10, pp. 1863– 1883, 2019.
[14] Y. Li, F.-X. Wu, and A. Ngom, “A review on machine learning principles for multi view biological data integration,” Briefings in Bioinformatics, vol. 19, pp. 325–340, 12 2016.
[15] C. Xu, D. Tao, and C. Xu, “Multi-view intact space learning,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 37, no. 12, pp. 2531–2544, 2015.
[16] C. Zhang, Y. Cui, Z. Han, J. T. Zhou, H. Fu, and Q. Hu, “Deep partial multi-view learning,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 44, no. 5, pp. 2402–2415, 2022.
[17] Y. Zhu, Q. Sheng, J. Cao, Q. Nan, K. Shu, M. Wu, J. Wang, and F. Zhuang, “Memory-guided multi-view multi-domain fake news detection,” IEEE Transac tions on Knowledge and Data Engineering, vol. 35, no. 7, pp. 7178–7191, 2023.
[18] X. Chen, F. Zhou, G. Trajcevski, and M. Bonsangue, “Multi-view learning with distinguishable feature fusion for rumor detection,” Knowledge-Based Systems, vol. 240, p. 108085, 2022.
[19] H. Phan, O. Y. Chén, M. C. Tran, P. Koch, A. Mertins, and M. De Vos, “Xsleepnet: Multi-view sequential model for automatic sleep staging,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 44, no. 9, pp. 5903–5915, 2022.
[20] H. Phan, H. Le Nguyen, O. Y. Chén, L. Pham, P. Koch, I. McLoughlin, and A. Mertins, “Multi-view audio and music classification,” in ICASSP 2021 - 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pp. 611–615, 2021.
[21] X. Sun, S. Sun, M. Yin, and H. Yang, “Hybrid neural conditional random fields for multi-view sequence labeling,” Knowledge-Based Systems, vol. 189, p. 105151, 2020.
[22] L. Zhuang, L. Wayne, S. Ya, and Z. Jun, “A robustly optimized BERT pre-training approach with post-training,” in Proceedings of the 20th Chinese National Confer ence on Computational Linguistics, (Huhhot, China), pp. 1218–1227, Chinese In formation Processing Society of China, Aug. 2021.