A new report which maps online antisemitism in the Netherlands, points to a structural problem across several platforms.
The report, authored by the Utrecht Data School on behalf of Central Jewish Consultative Committee (CJO) and the Information and Documentation Israel (CIDI), is the first structural exploration of the form and extent to which antisemitism is expressed online in the Netherlands.
Online antisemitism is a transnational problem. It is not clear exactly how many online antisemitic incidents and expressions occur. What is clear is that it is impossible to keep track manually – even if one limits oneself to Dutch (and Dutch-language) online antisemitism.
The report analysed cases of antisemitism in 2020 on a number of social media platforms, and a selection of websites and weblogs. This is the first application of a specially developed methodology to recognise antisemitic statements and is therefore a valuable first measurement.
Future iterations of the research will be necessary to gain insight into trends and the effect of measures against hate speech.
In order to estimate the problem reasonably, methods and algorithms had to be developed. Using search programs, Twitter, YouTube, Telegram, Reddit, a number of Dutch-language Facebook pages and 5,667 Dutch websites were searched for antisemitic statements.
In total, more than 1,800,000 messages were found in the course of the preparation of the report. These were further (automatically) analysed for actual antisemitic content by means of carefully ‘trained’ algorithms. Of these, some 200,000 were labelled as antisemitic.
Only antisemitic texts could be investigated. The available technology is not yet suitable for the analysis of images, sound fragments or videos. “Although a great deal of online antisemitism takes this form, antisemitic memes have unfortunately not been investigated,” explains CIDI director Hanna Luden.
The findings point to a large, structural problem with online antisemitism. On almost every platform, even on more strictly moderated platforms such as Twitter, much hatred against Jews was found. However, there is a visible difference with platforms where no effective moderation takes place.
A relatively large amount of antisemitism was observed on Telegram and YouTube in particular. The expressions are often more openly and extensively antisemitic. About 25% of all Dutch-language comments about Jews on YouTube turn out to be antisemitic in nature (the films themselves were not investigated, unfortunately). On Telegram this is 36%, and more than two out of three of these comments are explicitly antisemitic.
A first step has been taken, but there is still a long way to go. Research software must be developed further in order to also be able to search visual and audio material automatically. Reporting online antisemitism must be actively promoted, so that platforms are forced to react and take action. Reporting is also essential to understanding the nature and scale of the problem. Offline’ incidents of intimidation and violence are often fed online, so online hatred must be spotted and removed quickly.
CIDI director Hanna Luden: “We see that antisemitism online can quickly have physical consequences for people in the real world. More research and method development is needed to combat antisemitism effectively.”
The research was realised with a grant from the Ministry of Justice and Security. Advice on the follow-up of this research has been entrusted to the National Coordinator on Combating Antisemitism (NCAB).
At the same time, NewsCheckers and Textgain published a targeted study of political antisemitism on Twitter. Tweet behaviour of politicians from six different Dutch political parties was analysed. This showed that politicians from the FvD party by far the most often make mistakes and regularly retweet anti-Semitic accounts.
