Social Network and Topic Modeling Analysis of US Political Blogosphere

Transcription

1 Social Network and Topic Modeling Analysis of US Political Blogosphere Mark Burdick PhD Supervisors: Prof. Dr. Adalbert F.X. Wilhelm Dr. Jan Lorenz 1

2 Not the Research Question How do ideologies and social networks change together? 4

3 Locating the Project From Social Science ideology s role in textual content creation, in particular, political media. Social Science From Web Science - webcrawling, webscraping, and automated network mapping From Computational Analysis - semantic analysis techniques and visualization approaches Web Studies Computational Analysis 5

4 Ideology Rhetoric Teun van Dijk: Ideology: A Multidisciplinary Approach (1998), Politics, Ideology, and Discourse (2014), News As Discourse (2013) Connects ideology to discourse to rhetoric to news media Working definition based on van Dijk: Rhetoric is ideologically driven semantic avoidances or exaggerations achieved by deletion of words, substitutions of terms or names, and the repetition of certain topics. 6

5 - Webcrawling Webcrawling, Webscraping, Hyperlink Analysis - Webscraping - Hyperlink analysis - Influencer analysis (Bakshy, 2011) Influencers in a network (Seo, 2013) 7

6 - Comparative topic modeling (Alexander, 2016) Semantic Analysis and Visualization - Comparative word clouds (Diakopoulos, 2015) - Agent based modeling (Lorenz, 2011) An Example of a Comparative Word Cloud where pink represents mainstream media and blue represent blogs (Diakopoulos, 2015) 8

7 Why the US Political Blogosphere? - A social network consistently creating ideologically motived text - Fake News, Alternative Facts, Post-Truth - Blogs are increasing important for politics (Wallsten, 2007; Park, 2009) - Previous research on political blogs (Adamic & Glance, 2004; Diakopoulos et al, 2015) 10

8 Test Dataset Three blog posts from ten blogs every day for one month = approx. 850 documents Topic 1 Topic 2 Topic 3 Topic 4 Topic 5 Topic 6 Topic 7 Topic 8 Topic 9 Topic 10 tax trump obama clinton conservative political republican clinton trump clinton new clinton muslim govern people elect american clinton problem govern media war fbi class left trump trump debate bias million women isi campaign reason online democrat country poll mrs money campaign islam depart people good president people voter big Blogs from Blue Group Topic 1 Topic 2 Topic 3 Topic 4 Topic 5 Topic 6 Topic 7 Topic 8 Topic 9 Topic 10 political clinton pence trump trump white trump work clinton immigration people trump right tax campaign president women american trump crime american elect court pay republican war sexual job debate people elect voter support new president iraq assault security president black america vote issue return comment veteran woman social question hate Blogs from Red Group 11

9 Visualizations 18

10 - Blogs selected based on Alexa rating - 40 Blogs in three two-month batches - Batch dates selected based on approaching political events: - G20: July 7, Local Elections: November 7, Russian Election: March 11 or 18, 2018 Building the Database Data Collection Post ID Blog ID Post Title Post URL Day Post Text NRA strangely silenthttp:// 1 Wait. What s that noise? It Hillary Clinton's debhttp:// 1 Hillary Clinton and Donald T Donald Trump is sel 1 Donald Trump has made an Hillary Clinton Scramhttp:// 1 Democratic presidential no Mike Pence: Donaldhttp:// 1 Republican Vice Presidenti The Country Can t Suhttp:// 1 Supporting Donald Trump i Taking A Stand On S 1 AARP has asked all candida One More Time: Whhttp:// 1 Let me make it clear, I am n What To Read On Tr 1 I recently criticized US polit Trump: Blacks In Wohttp://talkingpointsmemo. 1 In an interview aired on Fri Trump: Clinton Has Ghttp://talkingpointsmemo. 1 During a campaign rally in A Clinton To Visit Charhttp://talkingpointsmemo. 1 Hillary Clinton's campaign c Intelligence Official 1 US intelligence officials are Trump Ohio Deputy 1 People across America reac Ted Cruz Endorses Thttp:// 1 For months, Ted Cruz has re Trump advisor: Evehttp://hotair.com/archives 1 The advisor is Omarosa Man Surprise! Team Trumhttp://hotair.com/archives 1 The spin: Sources from Don Inevitable: Ted Cruzhttp://hotair.com/archives 1 I predicted in July that a Cru Trump Earns Surprishttp:// 1 One of the most interesting Hillary Shouting Durhttp:// 1 An anesthesiologist (and m Woman Harasses Truhttp:// 1 Donald Trump drives libera Hillary's Pneumoniahttp:// 1 Only four days after I wrote Trump and the Anti- 1 If people call you an intell The Anger Is Real 1 This weekend, I talked with HILLARY: THE THIRD 1 On Monday, Donald Trump THERE S SOMETHINGhttp:// 1 Yesterday Tom Lifson took BASKET OF DEPLORhttp:// 1 David Brock, who runs neck Trump s new SCOTUhttps://thinkprogress.org/m 1 Last May, Republican presid Trump praises Egypthttps://thinkprogress.org/t 1 Just days after Republican p Hillary Clinton gets ahttps://thinkprogress.org/c 1 Since her campaign began, New York Times strohttp:// 2 With just over six weeks to Whoops! Hillary Clinhttp:// 2 Democratic presidential can URL, Text Body, Hyperlinks Politico: GOP Senatohttp:// 2 Establishment Republican s Jeff Sessions Pusheshttp:// 2 Alabama Senator Jeff Sessi Trump The Arsonist 2 Michelle Obama said on Se Webcrawling front pages Webscraping articles Table: Blog, Date, Title, 14

11 Data Analysis Analysis Technique Unit of Analysis Purpose of Analysis Longitudinal Social Network Analysis Influencer Analysis Topic Modeling Weekly hyperlinks of blogs Weekly and total hyperlinks of blogs Weekly and total text bodies from individual blogs and social groups Determine structure and changes of the blogosphere Determine importance of individual blogs within a network Enable computational comparison of semantic differences across blogs Rhetoric Identification Topic Models Reduce topic models to instances of substitution, repetition, and deletion (rhetoric) Predictive Modeling Hyperlinks and Rhetoric Validate relationship between rhetoric content and influencer status 16

12 Questions? Feedback? Responses?

13 References Adamic, L. A. & Glance, N. (2005). The Political Blogosphere and The 2004 US Election: Divided They Blog. In Proceedings of the 3rd International Workshop on Link Discovery (pp ). ACM. Alexander, E. & Gleicher, M. (2016). Task-driven Comparison of Topic Models. IEEE Transactions on Visualization and Computer Graphics. 22(1), pp IEEE. Bakshy, E., Hofman, J. M., Mason, W. A. & Watts, D. J. (2011). Everyone's An Influencer: Quantifying Influence on Twitter. In Proceedings of the Fourth ACM International Conference on Web Search and Data Mining (pp ). Van Dijk, T. A. (1997). What is Political Discourse Analysis. Belgian Journal of Linguistics. 11(1), pp Van Dijk, T. A. (1998). Ideology: A Multidisciplinary Approach. Sage. Newbury Park, CA. Van Dijk, T. A. (2013). News as Discourse. Routledge. New York. Seo, H., Kim, W., Lee, J., & Youn, B. (2013). Network-Based Approaches for Anticancer Therapy (Review). International journal of oncology. 43(6), pp Lorenz, J., Rauhut, H., Schweitzer, F. & Helbing, D. (2011). How Social Influence Can Undermine the Wisdom of Crowd Effect. Proceedings of the National Academy of Sciences. 108(22), pp

14 References Park, D. W. (2009). Blogging with Authority: Strategic Positioning in Political Blogs. International Journal of Communication. 3(24), Wallsten, K. (2007). Agenda Setting and the Blogosphere: An Analysis of the Relationship Between Mainstream Media and Political Blogs. Review of Policy Research. 24(6), pp Diakopoulos, N., Elgesem, D., Salway, A., Zhang, A. & Hofland, K. (2015). Compare Clouds: Visualizing Text Corpora to Compare Media Frames. In Proceedings of IUI Workshop on Visual Text Analytics. 4th Workshop on Visual Text Analytics, Atlanta, Georgia, USA.