Description

In this visualization, the sentiment of the articles of the official magazines by the IS, Dabiq and Rumiyah, are analysed. For each issue, all the articles are ordered chronologically. The height of the bar is determined by the length of the article measured in number of words. The color corresponds to the prevailing sentiment assigned by IBM’s Tone Analyzer after the extraction of text. To verify the accuracy of sentiment assignment, we manually checked a few random articles of the magazines and compared our results. The results were compatible so we felt confident enough to keep the results returned by the tool.

The two sentiments most present are sadness and fear. It is important to state that fear in this context it is not suffered by the IS but instead it is the fear that the IS itself wants to spread over the Western world.

Additionally, to compare the overall decreasing length of the text in the articles of the magazines (Note: the first six numbers of Dabiq have 5-7 articles, from issue #7 to #15 the number of articles doubles while the length of the magazine maintains the same number of pages) in a second analysis we considered also the number of images used in every edition of the magazines. What one can infer is that while the length of text decreased in time, the number of images used increased simultaneously. This could be in accord with the thesis of the IS being very much aware of the importance of media communication, treating visual content with privileged attention on its magazines.

Protocol

To analyze the brand image of ISIS, we proceeded by analyzing their official text production extracted from their digital magazines written with the English language: Dabiq and Rumiyah.

The magazines were obtained from the association The Clarion Project which gathers and collects different Islamic material. The article’s text was analyzed using IBM’s Tone Analyzer to extract the tone and sentiment of the text. The data obtained was later processed with Excel and OpenRefine to clean it and to facilitate the procedure of designing the visualisation. Also, we had to manually correct some translation errors or misunderstanding of the words by the software.

Data

Timestamp: 06/12/2016 - 13/12/2016

Data source: Tone analyser, The Clarion Project

Download data

The dataset provides three sheets: one per each magazine (Dabiq and Rumiyah) and one for total number of images. In the first two sheets, columns of data are magazine name, magazine issue edition id, article id, sentiment name, sentiment score, article length (count in keystroke). All sentiment scores were extracted using IBM’s Tone Analyzer. The sheet for data about images contains magazine issue id and total number of images per issue.