本篇論文提出一個強健語言模型加上額外特徵的系統，處理SemEval 2019
RumourEval: Determining rumour veracity and support for rumours (SemEval
2019 Task 7)，主要包含了兩個任務，任務A 為 使用者的立場偵測，任務B偵測
謠言是真、假或未驗證， 本研究利用到了對話分支的追蹤資訊，使用強健的預
訓練語言模型與詞頻特徵，加上報導其他特徵的深度學習預訓練模型，結合兩者
的預測結果，做出任務A的立場驗證，其Macro F1達到62%，再透過規則模型處
理任務B的消息驗證，達到 Macro F1 68%，且 RSME降到0.5983。

In this paper, we propose a robust language model with external features to
deal with SemEval 2019 RumourEval: Determining rumour veracity and support
for rumours (SemEval 2019 Task 7), which mainly contains two tasks. They are
Task A: User’s stance detection, and task B: detect whether the rumour is true,
false or unverified. We used the tracking of the dialogue branch, a robustly pretrained language model and word frequency features concatenate a deep learning
pre-trained model that reported other features. Concatenating the prediction results of the two, we reached the performance of 62% Macro F1 for task A , and
then processed the message verification of task B through a rule-based system to
reach Macro F1 68% where is RMSE is reduced to 0.5983.

[1] C.Castillo, M.Mendoza, and B.Poblete. Information credibility on twitter. In Proceedings of the 20th international conference on World wide web, pages 675–684,2011, ACM.
[2] G.Gorrell, E.Kochkina, M. Liakata, A.Aker, A.Zubiaga, K.Bontcheva, and L. Derczynski. SemEval-2019 task 7: RumourEval, determining rumour veracity and support for rumours In Proceedings of the 13th International Workshop on Semantic Evaluation, pages 845–854, Minneapolis, Minnesota, USA, June 2019. Association for Computational Linguistics.
[3] L.Derczynski, K.Bontcheva, M.Liakata, R.Procter, Geraldine.Hoi, and A.Zubiaga. Semeval-2017 task 8: Rumoureval: Determining rumour veracity and support for rumours. In Proceedings of the 11th International Workshop on Semantic Evaluation (SemEval-2017), pages 69–76,2017.
[4] E. J. L., Finding structure in time,” Cognitive Science, vol. 14, no. 2, pages 179–211,1990.
[5] S. Hochreiter and J. Schmidhuber, “Long short-term memory,” Computation, vol. 9, no. 8, pages 1735–1780, 1997.
[6] J. Gehring, M. Auli, D. Grangier, D. Yarats, and Y. N. Dauphin, “Convolutional sequence to sequence learning,” in Proceedings of the 34th International Conference on Machine Learning-Volume 70, pages 1243–1252, JMLR. org, 2017.
[7] A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, L. Kaiser, and I. Polosukhin, “Attention is all you need,” in Advances in neural information processing systems, pages 5998–6008, 2017.
[8] J.Turian, L.Ratinov, and Y.Bengio. Word representations: A simple and general method for semi-supervised learning. In of the 48th Annual Meeting of the Association for Computational Linguistics, pages 384–394,2010.Association for Computational Linguistics.
[9] A.Mnih and G.Hinton. A scalable hierarchical distributed language model. In Neural Information Processing Systems 21, pages 1081–1088,2008.
[10] R.Kiros, Y.Zhu, R. Salakhutdinov, R.Zemel, R.Urtasun, A.Torralba, and S.Fidler. Skip-thought vectors. In in neural information processing systems, pages 3294–3302,2015.
[11] M. Peters, M. Neumann, M. Iyyer, M. Gardner, C. Clark, K. Lee, and L. Zettlemoyer, Deep contextualized word representations, in Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 2227–2237, 2018.
[12] P. Rajpurkar, J. Zhang, K. Lopyrev, and P. Liang, “SQuAD: 100,000+ questions for machine comprehension of text,” in Proceedings of the 2016 Conference on Em- pirical Methods in Natural Language Processing, (Austin, Texas), pages 2383–2392, Association for Computational Linguistics, Nov. 2016.
[13] R. Socher, A. Perelygin, J. Wu, J. Chuang, C. D. Manning, A. Ng, and C. Potts, “Recursive deep models for semantic compositionality over a sentiment treebank,” in Proceedings of the 2013 Conference on Empirical Methods in Natural Language Processing, (Seattle, Washington, USA), pages 1631–1642, Association for Computa- tional Linguistics, Oct. 2013.
[14] E. F. Tjong Kim Sang and F. De Meulder, “Introduction to the conll-2003 shared task: Language-independent named entity recognition,” in Proceedings of the Sev- enth Conference on Natural Language Learning at HLT-NAACL 2003, CONLL ’03, (Stroudsburg, PA, USA), pages 142–147,2013. Association for Computational Linguistics.
[15] 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, 2019.
[16] S. R. Bowman, G. Angeli, C. Potts, and C. D. Manning, A large annotated corpus for learning natural language inference, in Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing, (Lisbon, Portugal), pages 632– 642,2015. Association for Computational Linguistics, Sept.
[17] A. Radford, K. Narasimhan, T. Salimans, and I. Sutskever, Improving language understanding by generative pre-training,2018.
[18] S.Mohammad, S.Kiritchenko, P.Sobhani, X.Zhu, and C.Cherry. Semeval-2016 task 6: Detecting stance in tweets. In Proceedings of the 10th International Workshop on Semantic Evaluation (SemEval-2016), pages 31–41,2016.
[19] E.Kochkina, M.Liakata, and I.Augenstein. Turing at semeval-2017 task 8: Sequential approach to rumour stance classification with branch-lstm. arXiv preprint arXiv:1704.07221,2017.
[20] T.Mikolov, K.Chen, G.Corrado, and J.Dean.Efficient estimation of word representations in vector space.arXiv preprint arXiv:1301.3781,2013.
[21] M.Fajcik, L.Burget and P.Smrz.BUT-FIT at SemEval-2019 Task 7: Determining the Rumour Stance with Pre-Trained Deep Bidirectional Transformers. In Proceedings of the 13th International Workshop on Semantic Evaluation 13 (2019), pages 1097-1104,2019.
[22] M.Peters, M.Neumann, M.Iyyer, M.Gardner, C.Clark, K.Lee, and L.Zettlemoyer. Deep contextualized word representations.preprint arXiv:1802.05365,2018.
[23] A.Radford, K.Narasimhan, T.Salimans, and I.Sutskever.Improving language
understanding with unsupervised learning. OpenAI,2018.
[24] Q.Li, Q.Zhang and L.Si.eventAI at SemEval-2019 Task 7: Rumor Detection on Social Media by Exploiting Content, User Credibility and Propagation Information.In Proceedings of the 13th International Workshop on Semantic Evaluation,pages 855–859,2019.
[25] Y. Liu, M.Ott, N.Goyal, J.Du, M.Joshi, D.Chen, O.Levy, M.Lewis, L.Zettlemoyer, and V.Stoyanov. Roberta: A robustly optimized bert pretraining approach.ArXiv, abs/1907.11692,2019.
[26] A.Prakash, H.T.Madabushi. Incorporating Count-Based Features into PreTrained Models for Improved Stance Detection .arXiv:2010.09078,2020.