For 20 years, ISD has delivered field-leading threat detection, analysis and real-world strategies to combat terrorism, extremism and authoritarianism - in all their ideological forms.

Home / Glossary

ISD Glossary

ISD’s Glossary provides clear definitions of key terms related to extremism, hate, information operations and tech accountability. This page is continually reviewed to ensure accuracy.

Terrorism  

The Global Internet Forum to Counter Terrorism noted in a 2021 report that there is “no universal agreement on the definition of terrorism.” In their report they list the different definitions used in countries like the UK, USA or Japan, regional bodies like the EU and UN, and technology companies like Facebook, Twitter and Microsoft. ISD does not have an in-house definition of terrorism, but instead use the definition of “terrorist acts,” provided by the UN, that describes them as “criminal acts, including against civilians, committed with the intent to cause death of serious bodily injury, or taking of hostages, with the purpose to provoke a state of terror in the general public or in a group of persons or particular persons, intimidate a population.”13 The use of politically or ideologically motivated violence can include promotion or support of groups associated with this violence, direct calls to violent action or the sharing of ideological material that may inspire others to carry out attacks.  

Transformer Models  

Transformer Models are a type of artificial neural network  that comprehend context and thereby grasp significance by observing associations in sequential information, such as the words in a text. Utilising a dynamic set of mathematical strategies, known as attention mechanisms, transformer models can discern the ways in which even separate elements within a data series impact and relate to one another. First introduced by Google in a 2017 paper, transformer models represent one of the most recent and potent models developed thus far, propelling a surge of breakthroughs in machine learning.