sebastianbarfort / sds

Social Data Science, course at University of Copenhagen
http://sebastianbarfort.github.io/sds/
12 stars 17 forks source link

Group 16 - Project description #77

Closed mertz2001 closed 8 years ago

mertz2001 commented 8 years ago

What makes news, news on social media

Recent studies by Ethan Zuckerman suggest that the digital media of today is less representative in a global sense than the old analogue media types and that social media is even less representative. Especially when considering the social media he explains how their targeting of stories suggested by algorithms to be relevant for the individual narrows our view on the global world since the media picture is focused on news close to the individual and his or hers beliefs. This along with previous results from the Reuters institute digital report (2012) implying that an increasing part of the younger generations in Britain gets the majority of their news stories from the social media is what motivates us to consider a similar investigation of the media in a Danish context. The purpose of this paper is twofold: First we wish to investigate the Danish media’s focus and check whether or not social media is in fact more focused on inland news than the official media websites. Since we cannot see exactly what type of news everyone gets expose to we do this comparing the same media on its official website and on their official Facebook page. This allow us to see if only a specific type of news gets into or left out of the social media platform. We compare the distribution of inland and foreign news to see how large a fraction of the Danish news concerning themselves with other parts of the world and how this differs between the official webpage and their Facebook page. In this sense we are able to see if the news media target their readers on social media narrower than their usual reader. Second we wish to consider the news ending up on Facebook and use number of likes, comments and times shared as a proxy for how many individuals getting exposed to the news and investigate if a specific type of Facebook news gets more exposure than others.
Data In this analysis we are going to use data from a number of the biggest news media in Denmark. Among the selected are Berlingske tidende and Danmarks Radio. We collect all articles published within the last year from two different platforms, the media’s official website and their corresponding Facebook page. The articles from the medias’ official websites will be scrapped by the use of a function looping through each article saving the section, title, summary, text and date. Posts from Facebook will be scrapped through Facebook’s API. Method We will analyze the words used in the articles to categorize which countries the articles is about, and if any words or categories determines if the article is posted on the media’s Facebook page. We will split the data in a training and a test dataset and make a regression to find out which articles theoretically would be posted on the media’s Facebook page. We will use the GLMNET approach. This will be compared to which articles actually were posted on the Facebook page. Literature: Sunstein, Cass. 2001. “Echo Chambers: Bush v. Gore, Impeachment, and Beyond” Zuckerman, Ethan. 2014. “Rewire: Digital Cosmopolitans in the Age of Connection”