The next speaker in this session at the SEASON 2025 conference is Aleksandra Urman, with a paper that audits the way the Russian-Ukrainian war has been framed in search results before and after the full-scale Russian invasion of Ukraine in 2022. Search engines are of course key pathways to the news, especially during crises, and are often trusted more than any other source of news (including mainstream media); this means that they have a considerable role to play in connecting information seekers to the news of the day.
The way search engines rank information, then, has the potential to affect people’s attitudes towards current issues and events; how this works, and how strong that effect is, remains contested, however. This might also be understood through the lens of framing theory, then; while this concept has traditionally been applied to the study of legacy media content, the design of search engine filtering and ranking processes can also be regarded as a form of framing, however. Search engines create a pattern of interpretation or representation which may elevate or discriminate against certain groups of people, or certain aspects of reality. This might in turn materially affect political processes across various countries.
In the present case, differences in framing might become more evident especially after the full-scale invasion of Ukraine; they should also be expected to be present in a comparison of Western (e.g. Google) and non-Western (e.g. Yandex) search engines, and of queries in different languages (e.g. English vs. Russian). The present project explored this through an algorithmic audit of search results over time, and this paper focusses especially on image search results in March 2021, June 2022, and January 2023 from Google and Yandex for the query ‘war Ukraine’ in English, Russian, and German as performed from a server in the Frankfurt region.
Images were chosen for this because they have a particularly high affective framing potential; selected images were coded manually before scaling this up through an automated approach, and coding focusses especially on locations, the presence of Ukrainian or Russian people, the apparent ahem gender, and number of individuals, their apparent roles (as military or civilians), etc.
There are a number of patterns here. Ukrainians are considerably more prominent throughout; this is especially pronounced on Yandex, in fact. Men are considerably more visible, and this is due especially also to a substantial representation of combatants and their equipment (especially also tanks) in images. Google queries in Russian show more civilians than queries in English or German; Yandex queries from 2022 show more ruins.
Overall, this points to some minor changes in representation from 2021 to later years, with only limited differences between Yandex and Western search engines. Yandex is known to have engaged in some explicit censorship especially of Russian war crimes in Ukraine, but the generic search terms used here might not have triggered such censorship. War is largely depicted as men with guns and tanks.
