dg club

“Russian activities such as troll farms, bot farms and deepfake are constantly evolving, reaching far beyond Ukraine. It affects public opinion in European countries," says Nina Khairova, Professor in Computational Linguistics at The Department of Computing Science, Umeå University.

ImageVictoria Skeidsvoll

Umeå University opened up to Ukrainian researchers when the war in Ukraine began. Nina Khairova, Professor in Computational Linguistics, is one of those who were welcomed. Here she is associated with the Humane AI Net, an EU-funded AI research programme with a total of SEK 500 million, of which Umeå University is a partner. She leads and participates in two studies, both aimed at using AI to address misinformation and disinformation detection regarding the war and its consequences.

Undermine citizens' trust

The first project studies the use of social media by public authorities. These accounts contribute to more transparency for citizens and have a high level of credibility as the public sector is obliged to comply with laws and regulations. However, if the laws are not followed, if the communication channel is updated in advance, or if users' comments are incompletely answered, problems can arise. “People then turn to online information, and these sources of information can be traditional news agencies, opinion leaders or 'ordinary' users. Migration issues are among the most debated on social media and can be particularly vulnerable to attempts of disinformation," says Professor Khairova.

The spread of misinformation can undermine citizens' trust and lead to significant social consequences, such as social inequality, the creation of social gaps and the manipulation of opinions. “Russian activities such as troll farms, bot farms and deepfake are constantly evolving, reaching far beyond Ukraine. It affects public opinion in European countries," Khairova says.

Advanced AI Methods

In collaboration with other European researchers, she addresses the misinformation detection in news articles problem and develops a solution based on a combination of AI algorithms such as supervised machine learning, text analysis and event argument extraction. With these advanced AI methods, it is possible to 'label' credible sources and less reliable sources, distinguishing which ones are more likely to be false. An extensive review of articles from reputable media in different regions across the world, including Europe, the US, Ukraine and Russia, has been carried out. Now, we are applying our approach to detect misinformation in news articles related to the Russian invasion of Ukraine published from February to September 2022 “Now, we are planning to apply our methodology for analysing textual information in Polish, Ukrainian and English, related thematically to issues of Ukraine and migration,” Khairova says.

Improve policies

In the second project, analysis-driven methods are used to extract deep insights from freely available text data on social media. "We want to offer solutions to help improve programmes and policies to foster a supportive environment for effective integration of Ukrainian migrants in their host countries," says Nina Khairova.
She has now been living in Umeå for 2 years and is learning Swedish. She feels both welcome and supported by her colleagues "up in the north."

"I am grateful to Umeå University and the Department of Computing Science. They have given me the chance to be part of interesting and useful research," Nina Khairova says.

Further information

Read more on the projects "Building Trust in Social Media: Assessing the Trustworthiness of Public Information on Ukrainian Migration to Europe" , and "Human-machine Collaboration for Content Analysis in the Context of the War in Ukraine" .