Proposal by Ritumbra Manuvie awarded

From the University of Groningen (UG) ICT multi-annual plan 2016-2020, a total of fourteen project proposals have been awarded to researchers at the UG and the University Medical Center Groningen (UMCG).

In the 'Call for Proposals Visualisatie, Geo en Data Science 2021' CIT invited researchers to submit innovative project proposals that focus on visualization, geo and data science, which can make an important contribution to science.

Award

Researchers of the awarded projects will receive a material budget of a maximum of € 3,000 and a maximum of 400 hours of support from CIT experts in the field of XR, data science geo data science and visualization.

Sense of Law

The interdisciplinary project proposal by Dr. Ritumbra Manuvie and Prof. dr. Jeanne Pia Mifsud Bonnici was awarded. Congratulations!

About the project

With the help of CIT, the group will aim to train AI models to identify patterns of online conspiratorial behaviour that are in conflict with the law.

We believe that once the algorithm is able to recognize 'illegal' and 'un-civil' behaviour we can build strategies to make online spaces safe and informative.

Background

In 2021 the team led by Ritumbra and Jeanne conducted a digital ethnography on the use of social media by political actors. Within the context of this research, they came with a large volume of anonymized textual data which can be identified as a predictor of ‘social and political unrest’ and can be classified into differential political-speech categories. This data can also be used by a political scientist to code statements (chat-instances) that are (or might be) in conflict with the law. They witnessed that often such online ‘legally conflicting’ statements translate into an offline conflict.

Close to home

For example, in May 2021, certain members of a Facebook community decided to host a protest in Groningen and assembled at a pre-announced time in the Noorderplatzoon park. The resulting assembly of people was found illegal by the Groningen police and those assembled were arrested. The same assembly would have been legal if people assembling would have notified the Groningen officials five days in advance of their intention to protest.

Several such instances repeating in different parts of the Netherlands made us think if there is a way through which simple information agents which can provide a Sense of Law (or rather a sense of legal procedures) can be built for the virtual space to avoid conflict with the law.

Use of the award

As a result of the award, the CIT data science team will assist in training classifiers that recognize when a chat is relevant to the topic of protest. Running this classifier on the available chats will enable the group to collect many examples of such lines, which they can then categorise further by a second model as being in conflict or not with the law (e.g. the 5 days advance notice of protest).

Goal

We want to demonstrate a proof-of-concept where such deep learning models can identify and categorize chat instances that fall under the particular legal use-case and can correctly augment or supply relevant legal information from the corpus of Laws relating to protest (as a test-case). The development of the test case will help us understand how legal information can be automatically supplied in a digital social media scenario, what are the possibilities and scalable scenarios.

The project will also give the team a jump-start toward the NWO grant in Digital Humanities and VENI grant during the 2022 application round. Good luck!