PRIO Senior Researcher Jonas Vestby has been awarded funding from the European Research Council for a groundbreaking study on how artificial intelligence and knowledge graphs can transform the way peace and conflict data is connected, understood and used.
Peace and conflict researchers have spent decades building datasets about wars, armed groups, political movements, ceasefires and other aspects of conflict. A new project, BRIDGE, will develop new ways of connecting this wealth of information – making it easier to understand how different sources relate to each other and where important differences lie.
Researchers studying conflict often combine several datasets to answer a question. But this can be more complicated than it appears. Two datasets may use the same name for an armed group or political actor while defining it in different ways. They may also disagree about when a group existed, who belonged to it, or whether two organisations should be treated as the same actor. These differences can easily disappear when datasets are simply linked using identification numbers.
“We have an enormous amount of valuable knowledge in peace and conflict research, but it has been collected by different projects, at different times and for different purposes,” says Jonas Vestby, PRIO Senior Researcher and project lead for BRIDGE. “BRIDGE is about connecting that knowledge without pretending that all the sources describe the world in exactly the same way. The differences between them can tell us something important too.”
Moving beyond rows and columns
Most conflict datasets are organised as tables, much like spreadsheets. Before researchers begin collecting information, they have to decide which categories and columns the dataset will contain.
But real conflicts rarely fit neatly into predefined boxes.
An armed group, for example, might split into several organisations, merge with another group, change its name or have unclear membership. Researchers collecting the data may describe these complications in written notes because there is no suitable place for them in the table.
BRIDGE will use knowledge graphs to capture more of this complexity. Instead of forcing information into fixed rows and columns, a knowledge graph can represent connections between people, organisations, events and other entities, while also recording where the information came from.
The project will convert existing conflict datasets into knowledge graphs and explore how information currently hidden in researchers’ notes can be incorporated.
Connecting numbers and detailed conflict research
BRIDGE will also look beyond quantitative datasets. Peace and conflict researchers have produced extensive qualitative research that examines individual conflicts in depth. Such studies can contain important information and nuances that are difficult to capture in conventional datasets.
BRIDGE will explore how AI and language models can help identify structured information in these texts and connect it with existing conflict data. This will allow researchers to examine where detailed studies and large datasets describe conflicts in similar ways – and where they see them differently.
The aim is not to use AI to decide which description is correct. Instead, BRIDGE will make it easier to trace information back to its original source and understand how different researchers arrived at their descriptions.
Building more transparent foundations for AI
The project comes at a time when generative AI and language models are increasingly being used to collect, organise and analyse political information.
This makes reliable and transparent research data particularly important. AI systems can process enormous amounts of information, but users also need to know where that information comes from and how the concepts behind it have been defined. BRIDGE will contribute to the development of open, AI-driven research tools built around these principles.
“As AI makes it easier to combine information from many different sources, it becomes even more important to know where that information came from and what it actually means,” says Jonas. “We want to build tools that preserve those connections, so that complexity and disagreement are visible rather than lost along the way.”
By connecting existing datasets with detailed conflict research, BRIDGE aims to make decades of accumulated knowledge easier to navigate, compare and use – while keeping its origins and complexities visible.
