Furthermore, hate speech is increasingly used by certain politicians, normalising its use enabled by the very contentious judicial standing of the status of hate speech, as there is a large gap between policy implementation and policy enforcement. Additionally, online hate becomes increasingly part of a new reality with Elon Musk’s ever growing disdain towards protections against hate speech. Consequently, the need for WhoDis as a tool to be able to track the links between the origin and dissemination of hate speech has never been so indispensable, especially as the more rooted the presence of online hate is as part of a new reality, the more progressive actors are demotivated.
However, WhoDis is not new!
On April 25, 2023, Justice for Prosperity published its WhoDis Report, which was launched at the European Parliament.
You can read the report below:
Within the WhoDis report, published by Justice for Prosperity in April 2023, the report outlined the 8 different modi-operandi used by subversive actors to spread their ideology.
These include:
- Networking
- Discourse
- Training
- Social Media/ Internet
- Funding
- Infiltration of influential positions
- Attacks on right defenders
- Reputational harm
What Justice for Prosperity has envisaged with the development of the WhoDis platform, as well as the WhoDis AI Visualisation tool, is the ability to chronologically examine and observe the development and the root cause of the modi-operandi of the divisive and hateful rhetoric that will shape the greater sphere of political discourse across wider society. Through the use of the WhoDis AI Visualisation tool, we found that the narratives and tactics used by the anti-rights movement are usually misleading and not transparent at all. Therefore, we have developed a measure of registering how polarising certain messages are interpreted.
The development of the WhoDis AI Platform aims to equip researchers, journalists, and policymakers with the tools necessary to identify and counteract online hate and polarisation effectively. The WhoDis platform and project is enabled by its four core characteristics : Analysis, Partnership, Innovation and Detection.
Analysis: Working together with world-leading journalists, academics and policymakers. Within these partnerships, JfP has been able to uncover different patterns and trends behind hate speech and how these manifest into larger scale movements. The analysis of these trends are based on a lexicon of keywords and strategic language used by anti-democratic actors as well as a taxonomy of anti-democratic actors or activities, digital data or information sources, and strategic language used. Once these trends are observed through the Lexicon and the taxonomy it permits identifying them as risk indicators.
Partnership: A core element of JfP’s work within the WhoDis project is its partnerships, together with different CSOs, Governmental Organisations, as well as marginalised communities, to together be aware, prepared and one step ahead of anti-rights defenders, as well as subversive actors and the threats they pose. These partnerships entail producing data and research valuable to the development of the project and detection of risk indicators An example of partnerships within the WhoDis Project can be seen through JfP’s collaboration with the Council of Europe which led to the collaboration with the CSO “No Hate Speech Movement Italia”, to examine the data results of their counter narrative campaigns aiming at undermining hate speech, specifically the campaign concerning that of the “Safer Internet Day” which started from February 6, 2024. Using the WhoDis Visualisation tool, JfP has assisted in helping No Hate Speech Movement Italia understand the spread, dissemination and reaction of specific keywords and hashtags within their campaigns against Hate Speech in Italy.
Innovation : Justice for Prosperity prides itself upon development and innovation. Based on the latest developments in AI, WhoDis operates as a tool that detects how external factors influence the spread of anti-rights activities, both online & offline by detecting the chronology of peaks of usage of hateful lexicon or taxonomy and connecting them to the offline usage of these terms. This allows to expose which offline situation caused these trends in lexicon usage and vice versa.
Detection: Identifying and exposing different actors responsible for the dissemination of hate speech, polarisation, and conspiracy theories and the implications this spread has upon our democracy, both online and offline. What makes the WhoDis Visualisation tool unique is the ability to use the foundation of Natural Language Processing innovation for the greater good. Indeed, Large Language Models refuse to process toxic language making it complicated for them to process hate speech. The rare previous platforms that worked on hate speech detection failed to be able to identify implicit hate speech messages unlike the WhoDis tool which detects this implicit language through identifying the level of positive and negative polarity that each message carries, as often hate speech is engineered to subversively appear as a ‘false positive’, using positive and affirmative language to disguise a more sinister meaning and implication, as well as being able to avoid certain trackers programmed to detect a limited lexicon which is not implicit with its intended meaning.
