Description

The visualization describes the predominant color models in the official magazine by the Islamic state: Dabiq and Rumiyah. The color analysis performed on every image has the aim to investigate the presence of a coordinated image behind the communication strategy behind the Islamic state.

Each row identifies an entire magazine edition. The colors are the ones that the Clarifai APIs evaluate as representative for every image analyzed. The colors are sorted by HEX values.

It is possible to have an overview of the color models in their normalized form for an easy visual comparison between the predominant colors for each edition of the magazine. Alternatively, the user can view the same color models weighted on their density (percentage of presence in the image from where the color model was extracted) as this is an effective reading that makes it possible to understand which is the actual chromatic predominance found in every edition of the magazines that the Clarifai APIs reveals from each image.

Protocol

The images from the official magazines by ISIS were analyzed. The available issues of Dabiq and Rumiyah were obtained from the association The Clarion Project which gathers and collects different Islamic material. Dabiq is composed of 15 issues, Rumiyah of 2 issues. The images were extracted from the PDFs using SmallPDF, an online tool, resulting in 1131 images for Dabiq and 66 for Rumiyah, in total 1197 images were obtained. In order to access the Clarifai API, the URLs for every image were needed. All the images were uploaded in edition-structured folders on GitHub and with the Google Chrome extension Web Scraper all the URLs were collected into an excel file. The urls were then used in batches of 20 on Clarifai to extract AI tags in 60 iterations. The resulting JSON files were mounted into one excel file using OpenRefine. The visualization was done in NodeBox.

Data

Timestamp: 06/12/2016 - 16/12/2016

Data source: Clarifai, The Clarion Project

The dataset has one column that identifies the edition of the magazine, one column has the URLs of the images, hence the images file names for identification, and columns for hex color codes, w3c hex color codes, w3c color names, and finally the color density. The dataset has a single page dedicated per each edition and one page with all the data listed together.