Topics: DATA

Jannick Schou and Johan Farkas at DataDrivenJournalism: ’Fake news’ has recently become a seemingly ubiquitous concept among journalists, researchers, and citizens alike. With the rise of platforms such as Facebook and Twitter, it has become possible to spread deliberate forms of misinformation in hitherto unforeseen ways. This has also spilled over into the political domain, where new forms of (disguised) propaganda and false information have recently begun to emerge. These new forms of propaganda have very real effects: they serve to obstruct political decision-making processes, instil false narratives within the general public, and add fuel to already heated sites of political conflict. They represent a genuine democratic problem.

Yet, so far, both critical researchers and journalists have faced a number of issues and challenges when attempting to understand these new forms of political propaganda. Simply put: when it comes to disguised propaganda and social media, we know very little about the actual mechanisms through which such content is produced, disseminated, and negotiated. One of the key explanations for this might be that fake profiles and disguised political agendas are incredibly difficult to study. They present a serious methodological challenge. This is not only due to their highly ephemeral nature, with Facebook pages being able to vanish after only a few days or hours, but also because of the anonymity of its producers. Often, we simply do not know who is disseminating what and with what purpose. This makes it difficult for us to understand and research exactly what is going on.

This post takes its point of departure from a new article published in the international academic journal New Media & Society. Based on the research done for this article, we want to offer some methodological reflections as to how disguised propaganda might be investigated. How can we research fake and disguised political agendas? And what methodological tools do we have at our disposal?…

two main methodological advices spring to mind. First of all: collect as much data as you can in as many ways as possible. Make screenshots, take detailed written observations, use data scraping, and (if possible) participate in citizen groups. One of the most valuable resources we had at our disposal was the set of heterogeneous data we collected from each page. Using this allowed us to carefully dissect and retrace the complex set of practices involved in each page long after they were gone. While we certainly tried to be as systematic in our data collection as possible, we also had to use every tool at our disposal. And we had to constantly be on our toes. As soon as a page emerged, we were there: ready to write down notes and collect data.

Second: be willing to participate and collaborate. Our research showcases the immense potential in researchers (and journalists) actively collaborating with citizen groups and grassroots movements. Using the collective insights and attention of this group allowed us to quickly find and track down pages. It gave us renewed methodological strength. Collaborating across otherwise closed boundaries between research and journalism opens up new avenues for deeper and more detailed insights….(More)”

Wasim Ahmed at Data Driven Journalism: “Journalists may wish to use data from social media platforms in order to provide greater insight and context to a news story. For example, journalists may wish to examine the contagion of hashtags and whether they are capable of achieving political or social change. Moreover, newsrooms may also wish to tap into social media posts during unfolding crisis events. For example, to find out who tweeted about a crisis event first, and to empirically examine the impact of social media.

Furthermore, Twitter users and accounts such as WikiLeaks may operate outside the constraints of traditional journalism, and therefore it becomes important to have tools and mechanisms in place in order to examine these kinds of influential users. For example, it was found that those who were backing Marine Le Pen on Twitter could have been users who had an affinity to Donald Trump.

There remains a number of different methods for analysing social media data. Take text analytics, for example, which can include using sentiment analysis to place bulk social media posts into categories of a particular feeling, such as positive, negative, or neutral. Or machine learning, which can automatically assign social media posts to a number of different topics.

There are other methods such as social network analysis, which examines online communities and the relationships between them. A number of qualitative methodologies also exist, such as content analysis and thematic analysis, which can be used to manually label social media posts. From a journalistic perspective, network analysis may be of importance initially via tools such as NodeXL. This is because it can quickly provide an overview of influential Twitter users alongside a topic overview.

From an industry standpoint, there has been much focus on gaining insight into users’ personalities, through services such as IBM Watson’s Personality Insights service. This uses linguistic analytics to derive intrinsic personality insights, such as emotions like anxiety, self-consciousness, and depression. This information can then be used by marketers to target certain products; for example, anti-anxiety medication to users who are more anxious…(An overview of tools for 2017).”