新聞媒體應傳遞公正且無偏頗的資訊，然而，現今許多新聞報導帶有特定的立場與政治傾向，這將會阻礙閱聽人獲得正確公正的資訊、也進而影響社會上偏頗觀點的形成。此研究的目的是辨識新聞內容中潛在的政治偏見，主要專注於台灣政治光譜的藍綠極端新聞分類。因受限於新聞內容無現有的政治立場標注、且人工標注繁瑣費時，我們提出了「政治傾向值」將新聞來源以立場分群，有效地解決這個問題。而現有對於新聞內容的政治偏見研究大多透過分析文中的引用句，但這類方法需依賴其他外部資源。我們透過提取同類的用字及句法結構為特徵，有效地分類偏藍及偏綠媒體。此外，此研究所提出的時間通用度能幫助提取出通用於不同新聞事件的特徵。透過實驗結果證實，我們提出的方法在分類準確率上超越其他模型，且提取之特徵用字、句法結構具有通用性和可解釋性。

Media outlets should present fair and accurate news in an unbiased fashion, however, nowadays many media outlets have certain political ideology and stance. This might cause readers unable to access impartial information, and biased public opinions can be formed. The goal of this research is to detect political ideology bias from news articles, we focus on two extreme sides of Taiwan's political spectrum: Pan-blue, Pan-green. One of the challenge is that there is no ideology bias label for news articles existed and manual labeling is time consuming. Our proposed `Political Ideology Rate' can evaluate and cluster news sources into certain classes to solve the problem. Our proposed method utilize clustered keywords and syntactic patterns as features to effectively classify news articles into two political ideology classes. Besides, the designed time general degree helps to extract event independent features. The experiments shows that our method outperform other baseline models, and the extracted features are generalizable and explainable.

Abstract
致謝
1 Introduction ——— 1
2 Related Work ——— 4
2.1 Media Bias ——— 4
2.1.1 Quote-based Method ——— 5
2.1.2 Content-based Method ——— 6
2.2 Political Ideology Detection ——— 6
3 Methodology ——— 8
3.1 Dataset Construction ——— 8
3.1.1 Dataset Collection ——— 8
3.1.2 Data Pre-processing ——— 9
3.1.3 Definition of Political Ideology Rate ——— 9
3.2 Pattern Extraction ——— 13
3.3 Graph Construction ——— 15
3.4 Token Categorization ——— 16
3.5 Topical Token Clustering ——— 19
3.5.1 Token Embedding ——— 19
3.5.2 Token Clustering ——— 19
3.6 Feature Weighting ——— 22
3.7 Classification Model ——— 24
4 Experiment & Results ——— 26
4.1 Baselines ——— 27
4.2 Feature Comparison ——— 29
4.3 Feature Analysis ——— 30
4.4 Political Bias of Other News Sources ——— 33
5 Conclusions and Future Work ———36
References ———38

[1] Hunt Allcott and Matthew Gentzkow. Social media and fake news in the 2016 elec-
tion. Journal of Economic Perspectives, 31:211–236, 05 2017.
[2] Alexandre Bovet and Hern ́an A. Makse. Influence of fake news in twitter during the
2016 US presidential election. CoRR, abs/1803.08491, 2018.
[3] Marta R. Costa-juss`a and Jos ́e A. R. Fonollosa. Character-based neural machine trans-lation. CoRR, abs/1603.00810, 2016.
[4] Chuanhai Dong, Jiajun Zhang, Chengqing Zong, Masanori Hattori, and Hui Di.
Character-based lstm-crf with radical-level features for chinese named entity recognition. In Chin-Yew Lin, Nianwen Xue, Dongyan Zhao, Xuanjing Huang, and Yansong Feng, editors, Natural Language Understanding and Intelligent Applications, pages 239–250, Cham, 2016. Springer International Publishing.
[5] Matthew Gentzkow and Jesse M Shapiro. What drives media slant? evidence from us daily newspapers. Econometrica, 78(1):35–71, 2010.
[6] Tim Groseclose and Jeffrey Milyo. A Measure of Media Bias*. The Quarterly Journal of Economics, 120(4):1191–1237, 11 2005.
[7] Yu-Lun Hsieh, Yung-Chun Chang, Yi-Jie Huang, Shu-Hao Yeh, Chun-Hung Chen, and Wen-Lian Hsu. Monpa: Multi-objective named-entity and part-of-speech annotator for Chinese using recurrent neural network. In Proceedings of the Eighth International Joint Conference on Natural Language Processing (Volume 2: Short Papers), Taipei, Taiwan, 2017. Asian Federation of Natural Language Processing.
[8] Mohit Iyyer, Peter Enns, Jordan Boyd-Graber, and Philip Resnik. Political ideology detection using recursive neural networks. In Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 1113–1122, Baltimore, Maryland, June 2014. Association for Computational Linguistics.
[9] Vivek Kulkarni, Junting Ye, Steven Skiena, and William Yang Wang. Multi-view models for political ideology detection of news articles. CoRR, abs/1809.03485, 2018.
[10] Michael Laver, Kenneth Benoit, and John Garry. Extracting policy positions from political texts using words as data. American Political Science Review, 97(2):311–331, 5 2003.
[11] Konstantina Lazaridou and Ralf Krestel. Identifying political bias in news articles. Bulletin of the IEEE TCDL
, 12, 2016.
[12] Yu-Ru Lin, James P. Bagrow, and David Lazer. “quantifying bias in social and mainstream media”by yu-ru lin, james p. bagrow, and david lazer with china-manau yeung as coordinator. SIGWEB Newsl.,(Summer), July 2012.
[13] Wang Ling, Isabel Trancoso, Chris Dyer, and Alan W. Black. Character-based neural machine translation. CoRR, abs/1511.04586, 2015.
[14] Tony Mullen and Robert Malouf. A preliminary investigation into sentiment analysis of informal political discourse. pages 159–162, 01 2006.
[15] Vlad Niculae, Caroline Suen, Justine Zhang, Cristian Danescu-Niculescu-Mizil, and Jure Leskovec. Quotus: The structure of political media coverage as revealed by quoting patterns. In Proceedings of the 24th International Conference on World Wide Web, WWW’15, page 798–808, Republic and Canton of Geneva, CHE, 2015. International World Wide Web Conferences Steering Committee.
[16] Eli Pariser. The filter bubble: What the internet is hiding from you. 2011.
[17] Daniel Preotiuc-Pietro, Ye Liu, Daniel Hopkins, and Lyle H. Ungar. Beyond binary labels: Political ideology prediction of twitter users. In ACL, 2017.
[18] RadimˇReh ̊uˇrek and Petr Sojka. Software Framework for Topic Modelling with Large Corpora. In Proceedings of the LREC 2010 Workshop on New Challenges for NLP Frameworks, pages 45–50, Valletta, Malta, May 2010. ELRA.
[19] Filipe Nunes Ribeiro, Lucas Henrique, Fabrcio Benevenuto, Abhijnan Chakraborty, Juhi Kulshrestha, Mahmoudreza Babaei, and Krishna P. Gummadi. Media bias monitor: Quantifying biases of social media news outlets at large-scale. In International Conference on Weblogs and Social Media, pages 290–299, 2018.
[20] Elvis Saravia, Hsien-Chi Toby Liu, Yen-Hao Huang, Junlin Wu, and Yi-Shin Chen. Carer: Contextualized affect representations for emotion recognition. In Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, pages 3687–3697, 2018.
[21] Yanchuan Sim, Brice D. L. Acree, Justin H. Gross, and Noah A. Smith. Measuring ideological proportions in political speeches. In Proceedings of the 2013 Conference on Empirical Methods in Natural Language Processing, pages 91–101, Seattle, Washington, USA, October 2013. Association for Computational Linguistics.