Investigative Data Journalism: Uncovering Opaque Relationships through Social Network Analysis

Social Network Analysis. Image: Public Domain
Social Network Analysis. Image: Public Domain

Two million globally active actors from politics, business, and other sectors of society are recorded in a database and can be examined for their relationships using the research tool Influence Networks. Data journalist Nicolas Kayser-Bril helped develop it. In this article, he shows how it works. In the video statement, he explains why journalists today need to learn from programmers.

In my opinion, journalists today need to become more like programmers. We see today that hackers, programmers, and developers create applications and websites that help transform information into meaning – which is, after all, the job of journalists. The best examples of this new type of journalist are Adrian Holovaty and Julian Assange. Holovaty (born 1981) is a developer and journalist who created EveryBlock, a local microblogging platform that helps connect neighborhoods.

What Assange, in turn, has achieved with WikiLeaks, he has achieved primarily as a hacker and programmer. That’s why I think journalists absolutely must learn how to process data on the internet – if they don’t, developers will take matters into their own hands.

Organizing collaboration

I taught myself how to program. At first, I mainly programmed small applications. I then took these skills to the company OWNI. There, my job is to bring journalists, developers, and designers together so they can work on information collaboratively. In my work, I notice how it’s becoming increasingly important for journalists to act as project managers and bring different people and professions together.

To give an example: We carried out a major project focusing on real estate and developments in the housing market. For this, we had collected thousands of data points. We had to organize these in a database. To make them understandable for end users, we needed interface designers. To generate the data in the first place, we needed data miners and developers. Together with all these people, we were then able to establish connections between this data and create a reliable picture of how the housing and real estate market in France is developing.

You are currently viewing a placeholder content from Vimeo. To access the actual content, click the button below. Please note that doing so will share data with third-party providers.

More Information

This is a good example of how people with different skills come together to produce a story. As is well known, this is called data journalism these days. The interesting thing about data journalism is that you can create interfaces that allow amateurs or ordinary citizens to provide data and perhaps even help process it. This is called crowdsourcing.

Crowdsourcing can be very simple: You open the comments section and ask the community, “Okay, tell us your story!” But crowdsourcing can go even further. You can also organize crowdsourcing – enabling users to participate in validation processes. One example is the principle of peer review. We also used this type of crowdsourcing for our Influence Networks project.

The research tool Influence Networks

On the Influence Networks homepage, you can choose between three actions: Add a relationship, verify a relationship, or visualize a relationship. Here you’ll also find the link to the About page. Most importantly, however, this is where you can report bugs, view the source code, and contribute to the code.

So, if you decide to “add a relation” and land on the “add a relation” page (see image below), you can enter any type of relationship between two actors and then specify its type. We are still in the process of establishing an ontology for these relationships, for example, using a list of concepts that describe each relationship in more detail. This is how we created the relationship type “rivalry,” among others. Not very scientific, but very important when you’re examining a group of people based on their relationships.

In the next step, the relationship you entered is reviewed by other users of Influence Networks. The main goal here is to assess the credibility of this relationship. The results are ultimately presented on the visualization screen. Here, users can see what connects people to one another. For example, if you click on Muammar al-Gaddafi twice here, you’ll see that he was friends with Nicolas Sarkozy from 2007 to 2011 and that they subsequently became rivals.

This example illustrates how Influence Networks can be used as a research tool to practice a new form of investigative journalism. I would call this investigative data journalism.

Leave a Reply

Your email address will not be published. Required fields are marked *.

This site uses Akismet to reduce spam. Learn how your comment data is processed.