A political scientist by training, one of Insight’s newest members James Cross says his work has become more ‘computationally oriented’ in recent years as technological advances have made new research tools available to study political text at scale. He joined the Insight Research Ireland Centre this year after securing funding for two major projects in the emerging field of Computational Political Science.

The first – The SDG_EU Project: Tracking the influence of sustainable development goals on legislative negotiations in the European Union using precision and ambiguity – focuses on the EU policymaking process, the committees that drive this process and the nature of the policy debates that shape the resulting policy agreements.

The UN sustainable development goals (SDGs) are the most high-profile political commitment to address the challenges faced by humanity in the 21st century. To unveil the influence of the SDGs on EU policy, the project will consider recorded committee debate videos from the Council of Ministers and European Parliament and the policy agreements that these committees shape.

The nature of these data requires an interdisciplinary approach, integrating cutting-edge methods from data science and machine learning with theoretical insights and statistical methods from political science.

Cross and the research team aim to collect and process all relevant content relating to SDG discussions in EU committees and then employ large language models (LLMs) to extract a comprehensive empirical picture of how precision/vagueness in committee debates shapes how SDG commitments manifest in the resulting policy agreements.

‘Without precise formalised policy commitments, there is little chance that the SDGs will come to fruition,’ says Cross. ‘If successful, our project will refocus our attention on the origin of such agreements (the policymaking process) and put committee deliberations at the heart of our understanding of the SDG implementation process. The insights that result can inform attempts to ensure the SDGs have policy bite and increase the chances that we successfully address the global challenges they seek to confront.’

The second project is called ParliView – Bridging the information divide between the people and parliament. Insight at University College Dublin has teamed up with the University of Strathclyde and Transparency International (EU) to improve the transparency of European politics for citizens.

‘Parliaments are the cornerstone of modern democracies, yet citizens struggle to monitor their activities,’ says Cross. ‘Despite far-reaching open-access data policies, the overwhelming quantity of data and the unwieldy nature of user interfaces mean that citizens see little benefit from these efforts. The potential benefits of increased parliamentary transparency thus remain unrealised.’

This challenge is significant because trust in democratic institutions such as the European Parliament is low (only 50% of respondents were confident in the EP in 2022; Eurostat 2022). Citizens feel disconnected from the politicians who represent them, and the information environment through which they learn about parliament’s activities is flooded with poor quality, misleading content. Part of the solution to these challenges is to increase parliamentary transparency, but current efforts to increase transparency fall short as the information provided never reaches the citizens it is targeted at.

ParliView is an AI-powered transparency tool that aims to help citizens better understand a Parliament’s actions. Using generative AI techniques, it will transform information from unwieldy open-access parliamentary databases into concise, user-friendly responses to user queries. ParliView makes open-access data about parliamentary activity genuinely accessible to citizens in simple, understandable language, thus bridging the knowledge gap.

ParliView will allow users to ask questions about the activities of parliamentary representatives in natural language through an intuitive chat-based web interface. The ParliView backend processes queries to identify a set of relevant recent speeches from an up-to-date database of parliamentary activities. A task-specific large language model (LLM) then generates a natural-language response based on this information. ParliView will thus allow users to gain insight into how parliamentarians represent their interests in a user-friendly and interactive manner.

The platform is being designed for anyone that wants to use it but should have particular value for first time voters or those who might opt out of voting because they feel overwhelmed by information, or indeed, disinformation, says Cross.

‘There’s a gap between the European Parliament and the people, politics feels remote. Voters don’t know what their parliamentarians are doing. We’re trying to build a tool that will go some way to addressing that gap in understanding.

‘We are not in the business of giving voting advice. We want to help the voter to figure out what a sitting MP has done in their time in office.’

Work with user groups begins this autumn.

James Cross is Associate Professor in the School of Politics and International Relations at University College Dublin, and the Director and co-founder of the Connected_Politics Lab @ UCD.