引入辯論圖結構之說服力預測

Incorporating Argument Structure for Persuasiveness Prediction

黃國祐(Kuo-Yu Huang)

指導教授 ： 陳信希

國立臺灣大學/電機資訊學院/資訊網路與多媒體研究所/碩士(2020年)

https://doi.org/10.6342/NTU202002107

摘要

辯論圖架構闡明了主張與前提之間的關係。 但是，以前在其他辯論探勘中的研究並未在其模型中考慮這種關係。本論文提出了一種稱為異構辯論注意力網絡的神經網路。此模型透過使用圖神經網絡模型來學習辯論圖結構，已達成對一場辯論中論點與論點之間的關係更好的理解。藉由同時訓練說服力預測和立場預測並搭配兩階段抽樣訓練資料的方式，我們的模型在ChangeMyView（CMV）數據集中的說服力任務上實現了最好的結果。實驗結果表明，我們的圖架構使我們的模型能夠有效地匯集跨多個段落的信息、我們的立場分析任務使得我們的模型可以分辨不同回復的立場和我們的兩階段抽樣可以幫助我們的模型學到更好的特徵。
幫助我們的模型在說服力預測上做得更好。

關鍵字

自然語言處理 ； 辯論探勘 ； 說服力預測

並列摘要

Argument structure elaborates the relationship among claims and premises. However, previous works in argument mining do not consider this relationship in their architectures. To take argument structure information into account, this thesis proposes an approach to persuasiveness prediction with a novel graph-based neural network model, called heterogeneous argument attention network HARGAN.By jointly training on the persuasiveness and stance of the replies, our model achieves the state-of-the-art performance on the ChangeMyView (CMV) dataset for the persuasiveness prediction task. Experimental results show that the graph setting enables our model to aggregate information cross multiple paragraphs effectively. In the meanwhile, our stance prediction auxiliary task enable our model to recognizing the stance for each reply, and can help our model performs better on the persuasiveness prediction.

並列關鍵字

Natural Language Processing ； Argument Mining ； Persuasiveness Prediction

參考文獻

Yamen Ajjour, Wei-Fan Chen, Johannes Kiesel, Henning Wachsmuth, and Benno Stein. Unit segmentation of argumentative texts. In Proceedings of the 4th Workshop on Argument Mining, pages 118–128, Copenhagen, Denmark, September 2017. Association for Computational Linguistics. . URL https://www.aclweb.org/anthology/W17-5115.

David Atkinson, Kumar Bhargav Srinivasan, and Chenhao Tan. What gets echoed?understanding the “pointers” in explanations of persuasive arguments. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), pages 2911–2921, Hong Kong, China, November 2019. Association for Computational Linguistics. . URL https://www.aclweb.org/anthology/D19-1289.

Andrew Eliot Borthwick. A Maximum Entropy Approach to Named Entity Recognition. PhD thesis, USA, 1999. AAI9945252.

Tuhin Chakrabarty, Christopher Hidey, Smaranda Muresan, Kathy McKeown, and Alyssa Hwang. AMPERSAND: Argument mining for PERSuAsive oNline discussions. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), pages 2933–2943, Hong Kong, China, November 2019. Association for Computational Linguistics. . URL https://www.aclweb.org/anthology/D19-1291.

Djork-Arné Clevert, Thomas Unterthiner, and Sepp Hochreiter. Fast and accurate deep network learning by exponential linear units (elus). In Yoshua Bengio and Yann LeCun, editors, 4th International Conference on Learning Representations, ICLR 2016, San Juan, Puerto Rico, May 2-4, 2016, Conference Track Proceedings, 2016. URL http://arxiv.org/abs/1511.07289.