You are here: Foswiki>CounterJihadism Web>ProjectGroup8 (18 Feb 2013, SophieWaterloo)Edit Attach

Islam, Hate and the Mainstream Internet: A Cross-Cultural Study of Actors and Sentiment on Google

Team Members

Sophie Waterloo, Audrika Rakshit, Seah Kim, Mihaela Naftanaila, and Eva Anagnostaki

Introduction

Since the terrorist attacks in New York on September 11, 2001, negative depictions of Islam and Muslims have been appearing in the western media (Abdulla). At the same time, the digital environment has become a new communication tool for both jihadist movements and counter-jihadist movements (Bermingham et al). A number of individuals and groups are reported to have used the Internet for both planning attacks and for preparation, as well as for training practices on how to make bombs or for propagating the Jihadist ideology (Awan 76). Combined, the online sphere has led to a transformation in the way Muslims are perceived, while also furthering the practice of terrorism.

This extension of terrorism and radicalisation online is, in this sense, conceived as a process whereby individuals, through their online interactions and exposure to various types of Internet content, come to view violence as a legitimate method of solving social and political conflicts (Bermingham et al. 1). Online radicalisation is also one of the risks of the fragmenting characteristics of the Internet, according to Sunstein. The Internet provides the option to create new groups and connections by means of discussion groups or E-mail (Sunstein 57). The same goes for advocates of anti-Muslim groups, who primarily feel that the Islam religion is dangerous and fear the effects of the spreading of this religion (Islamaphobia). However, hate content regarding religion is not only available to advocates of Islam or those anti-Islam, but to everyone accessing the Internet. This begs the question of ready availability of hate content online: does it show up in a simple, unspecified Google search or does one have to dig deep?

Therefore, this study focuses on the position of anti-Muslim content in a series of local Google searches for unspecified keywords relating to Islam. The feature of local Google domains provides the opportunity to gain a richer and more balanced understanding of the Islam issue space developing online, and to reveal differences between cultures in their sentiment towards Islam. This study also serves to explore and highlight any existent dominant actors that may be providing content on the subject of Islam in the considered countries. Further, it also allows the anti or pro sentiments prevalent across the local Google domains to be viewed against and related to the position taken by the dominant actors, if any.

Research Questions

To explore what local Google domain queries on unspecified Islamic keywords reveal about the type of sources and entry-points that a person is provided with in the mainstream Internet search, we ask the following questions:
  • How close is extremist (hate) content when querying for unspecified keywords related to Islam on local Google domains for the considered countries?
  • What does a local Google search on unspecified keywords related to Islam reveal about the anti or pro sentiment for the considered countries?
  • Which actors are dominant in the distribution of online content on Islam for the considered countries, and how does this relate to the dominant sentiment on Islam content?

Methodology

Sample and material

To answer the three research questions, a content analysis was performed on the search results of a set of five queries within local Google domains of the following countries: Belgium, Netherlands, Germany, India, France, United Kingdom and United States. These countries were chosen for their already existing or developing contentious relationship with Islam, the Islamic community and/or the Muslim community. Search engines continue to function as a popular gateway into the World Wide Web, in particular Google. By using the dominant local search engine as the basis for research, a representative result can be produced to indicate what people in a specific national culture are being confronted with each time they type in a certain query. To get a sense of the sentiment surrounding the Islamic issue space and the proximity of extremist (hate) content in mainstream Internet search, the following unspecified keywords relating to Islam were chosen to be queried: Muslim, Islam, Burqa, Allah and Jihad. Before running the queries on each of the local Google domains, the browser settings were changed to exclude ‘Google Instant predictions’ and ‘Do not take into account my history of queries’ to avoid the occurrence of any personalization in the search results. Thereafter, the first 20 results of each query per local Google domain were extracted using the Harvester tool and analysed. This limitation to the first 20 results is based on the notion that these first results receive the most amount of attention (Guan and Cutrell 420; Joachims et al. 5). The first two results are generally clicked more often, and results on rank 3 to 5 are less so (Joachims et al. 5). However, when one starts scrolling down, rank appears to be less related to attention. Still, it is an indication that the first couple of results are more important than results that are ranked lower.

Procedure

After the links were selected, each was coded on the actor type of the source and the anti or pro sentiment towards Islam. The actors were coded according to the following categories: news site, person (individual Twitter or Linked In accounts), company and organization site (both profit and non-profit), governmental/political party site, educational site, community site, informational site, entertainment site, weblog, and social network sites (groups or general). The number of times the type of actors appeared per country was counted to provide information on the most dominant actors. These were then put into word clouds, colour coded by type of actor. The content was furthermore coded as anti-Muslim, pro-Muslim, neutral, or not related. Extremist content was highlighted individually. To quantify the search results, we ranked each link per query and country using inverse ranking, so as to get a cumulative value that could be grouped together to provide an indication of pro or anti sentiment per local Google domains. The link appearing first in the search results were given a value of 20, as this position is most valuable in Google. The links appearing on the twentieth position of the search results were given the lowest value of 1. To visualize the position of the different types of sentiment and position of the same content in the Google search results, a colour-coded scheme was made.

Findings

Data_Sprint_Sentiment3.jpg

The above visualization charts the content sentiment across the first 20 listings of each of the seven considered local Google search results. As is evident, the access to anti and pro Muslim content, and in some cases, extremist content, is quick and easy.
  • The Netherlands, Germany, United States and the United Kingdom show an equitable dispersion of anti-Muslim and pro-Muslim content in their respective top 20 search results. Instances of extremist links are present for each country, except for the United States where no extremist links were found for any of the considered queries.
  • India is notable for the appearance of two extremist anti-Muslim links for the query ‘Jihad’ despite the predominance of largely neutral and unrelated content in most of the considered query results.
  • Remarkably, Belgium indicates instances of decidedly anti-Muslim and pro-Muslim content within the first 20 search results. Based on the cumulative inverse ranking scores, it is clear to see that there are instances of almost equal degree of available Pro-muslim (198) and anti-Muslim (195) content. Also of note is the appearance of extremist pro-Muslim links for the queries ‘Moslim’ and ‘Allah’ in the light of the fact that all the terms queried were unspecified and neutral in nature. Extremist content doesn’t need to be scoped out, it is available for ready access.
  • France is of interest for its large clusters of pro-Muslim content. The calculated cumulative inverse ranking score of 446 for pro-Muslim content is significantly higher than all other type code categories, including 'Neutral', as seen in the chart below. However, what is perhaps also worth highlighting is the appearance of 5 anti-Muslim links in the second half of the total search results for the query ‘Burqa’ and the occurrence of 13 anti-Muslim links, including one extremist link, in the search results for the query ‘Jihad’. This may be said to directly echo and reflect a sustained general dissent and ideological tussle among the French public ever since France’s ban on burqas in 2011.

Inverse_Ranking_Image.jpg

The Actors visualization below reveals the dominant sources per considered country that actively contribute to the present online Islamic issue space. While the links available among the top 20 results are varied, they do allow for relatively easy entry points to either distinctly pro or anti Muslim content. It is worthwhile, therefore, to identify, categorize, and trace the actors involved in shaping and pushing the same content.

Actors_Edited_v2.jpg

  • In the case of Belgium, India, and France, 4 of the top 5 sources were news sites, which corresponds with our findings that the top contributing category for each of these three countries were news sources - either news portals or web versions of print papers. Though newspapers and news sites are expected to be largely neutral in their opinion, there were instances of bias in the links from the key actors: Belgium - nieuwsblad.be (anti), gva.be (anti), flw.ugent.be (pro); France - islammedia.free.fr (pro), leparisien.fr (anti); India - indianexpress.com (anti).
  • Wikipedia.org was identified as the key actor and top contributor in the search results on the local Google for Germany, the Netherlands, the United States, and the United Kingdom. The successive actors for Germany and the Netherlands are largely same in nature, made up of informational, news and entertainment sites. In the case of the US and the UK, it is interesting to note the position held by social sites such twitter.com, youtube.com, amazon.com and facebook.com in contributing to the Islamic issue space. It can be said that unspecified terms, such as those used in this study, are yet to yield or point towards contentious content in the local Google domains for these two countries.

Conclusion

A consideration of the question of Islam and hate, and its proximity to the lay Internet user, led to an inductive exploration and hunt for “extremist content” using unspecified keywords on the seven considered local Google domains. The study raised the question: how close is hate? Quite close, it seems. Within the first five search results, a user in any of the seven countries, is confronted with 2 links containing anti-muslim content on average. Given that the queries used in this study are intentionally neutral and free of bias, this number is remarkably revealing of the presumably slow and steady move towards the development of a more fierce counter-jihadist sentiment online, or an anti-muslim sentiment, at any rate.

While the visualizations and numbers indicate the presence of significantly neutral or unrelated content in the respective search results, the spread of hate and anti-sentiment across the search results cannot be denied. The neutral and unrelated content follow directly from the dominance of news sites, informational sites and entertainment sites across the board, however, it is also important to bear in mind that many other actors, such as company/organisation sites, community sites, and blogs also combine with the prior mentioned actors to contribute towards a markedly divided sentiment towards Islam, as evidenced through local Google search queries. This is not to disregard the high count of pro-Muslim content, as well as extremist pro-Muslim content, especially in the case of Belgium and Germany.

Further research

As an extension to this research, more specific and a wider net of queries may be crafted to serve as a foil for the findings of this study. Less neutral query terms may serve in proving the occurrence, or lack thereof, of specifically counter-jihadist content online. The path to such content, and the ease of access, could also be compared with the findings of the neutral queries of this study, to provide a more robust understanding of how close hate really is to each Internet search user. Also interesting would be to expand the number of countries considered in studying the issue of hate in relation to Islam and how readily accessible it is to the average Internet user.

References

Abdulla, Rasha. “Islam, Jihad, and Terrorism in Post-9/11 Arabic Discussion Boards.” Journal of Computer-Mediated Communication 12.3 (2007). 15 January 2013. <http://jcmc.indiana.edu/vol12/issue3/abdulla.html>.

Awan, Akil. “Radicalisation on the Internet? The Virtual Propagation of Jihadist Media and its Effects.” The RUSI Journal 152.3 (2007): 76-81.

Bermingham, Adam, et al. Combining Social Network Analysis and Sentiment Analysis to Explore the Potential for Online Radicalisation. Proceedings of the International Conference on Advances in Social Networks Analysis and Mining, July 2009, Dublin City University. Dublin: Centre for Sensor Web Technology, 2009. 231-236.

Guan, Zhiwei, and Edward Cutrell. An Eye Tracking Study of the Effect of Target Rank on Web Search. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, April 2007, San Jose. New York: ACM, 2007.

Joachims, Thorsten, et al. Accurately Interpreting Clickthrough Data as Implicit Feedback. Proceedings of the International ACM SIGIR Conference on Research and Development in Information Retrieval, August 2005, Salvador Brazil. New York: ACM, 2005. 154-161.

Sunstein, Cass. Republic.com. New Brunswick, NJ: Princeton University Press, 2000.
Topic attachments
I Attachment ActionSorted ascending Size Date Who Comment
Actors_Edited_v2.jpgjpg Actors_Edited_v2.jpg manage 580 K 22 Jan 2013 - 14:11 SeahKim  
Data_Sprint_Sentiment3.jpgjpg Data_Sprint_Sentiment3.jpg manage 359 K 18 Jan 2013 - 10:21 SeahKim  
Inverse_Ranking_Image.jpgjpg Inverse_Ranking_Image.jpg manage 181 K 18 Jan 2013 - 10:22 SeahKim  
Topic revision: r13 - 18 Feb 2013, SophieWaterloo
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Foswiki? Send feedback