Data Journalism Awards featured winner: Riot Rumours
The winners of the 2012 edition of the Data Journalism Awards were announced on 31 May during a ceremony held at the News World Summit in Paris. In this series of posts we will present each of the six winning projects to understand their relevance to the field of data journalism and provide an overview of the tools and methods used by participants.
The first of the six winners to be showcased here, the Riot Rumours project, achieved first place under the Data Visualisation and Storytelling, national/international category. To see the jury comments click here.
The project interface on the "Reading the Riots" website
An interdisciplinary team of academics and some advanced web technologies supported The Guardian's most ambitious visualisation yet - to illustrate how rumours spread on Twitter during the August 2011 riots in England. "The result is one of the most ambitious pieces we have ever built, both in terms of data analysis and dynamic graphics. The purpose of the project was to display how misinformation corrects itself in open, unregulated forums," explained Alastair Dant, lead interactive technologist for the Guardian. The team that turned the project vision into reality was the Guardian Interactive Team together with Rob Procter (University of Manchester, director of the Manchester eResearch Centre), Farida Vis (University of Leicester, lecturer in Media and Communication) and Alex Voss (St Andrews, School of Computer Science). The project took two months to complete.
"We analysed 2.57m riot-related tweets, provided to us directly by Twitter. The work was performed by the entire Guardian Interactive Team, plus three university staff and researchers to help filter through the tweets," said Dant. The skills required for the project were coding, research, visualisation and ability to identify rumour. The main tools employed were WebGL, HTML5 Canvas, Flash and Backbone.js.
The team explains: "Once the tweets clusters were identified, we developed a system to visualize their rise and fall over time. Sizing each tweet according to the influence of its author (determined by follower count), we added a decay function that would allow it to dissipate over time. We classified each tweet according to a 'common sense understanding' of its main role as a communicative act." After seeking advice from some experienced friends, the team chose an architecture that resembles "what you'd find under the hood of an arcade game." The data for each rumour is loaded from a JSON file and placed in a structured model. Every frame, the model is queried to find what's in play, a custom physics engine is updated to reflect the results and a renderer draws the current state to the screen.
"We chose an interactive visualisation. Some of the challenges we met include how to work out which rumours should be tracked and how to isolate tweets that relate to these rumours, visualising the 'flow' of a particular rumour as it took flight, providing a parameterised Levenshtein distance algorithm for finding all tweets within a certain 'distance' from each other in textual terms. We also traced the narrative arc of each story with a series of editorially curated panels, each of which cites a key tweet (or tweets) from within the life of the rumour."
Visualising the origin and diffusion of riots-related rumours on Twitter
The project became one of the most popular visualisations ever produced by The Guardian with thousands of retweets and hundreds of mentions all over the Internet. The project was used in policy discussions and directly combats the idea that Twitter caused the riots. For this reason the project was instrumental in illustrating that Twitter did not cause the riots.
Advice for aspiring data journalists
"Try and provide a helpful context for what may otherwise seem a somewhat abstract journey and learn from the successful aspects of gaming."
The Data Journalism Awards is a Global Editors Network initiative supported by Google and organized in collaboration with the European Journalism Centre. Please visit the Data Journalism Awards website for the full list of winners.