Characterizing Knowledge Manipulation in a Russian Wikipedia Fork

Mykola Trokhymovych, Oleksandr Kosovan, Nathan Forrester, Pablo Aragón, Diego Saez-Trumper, Ricardo Baeza-Yates · 2025

Read Characterizing Knowledge Manipulation in a Russian Wikipedia Fork by Mykola Trokhymovych, Oleksandr Kosovan, Nathan Forrester, Pablo Aragón, Diego Saez-Trumper, Ricardo Baeza-Yates (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-10663.txt.

Introduction

When Russia's rules tightened, a new site called Ruwiki copied Russian Wikipedia and rewrote parts of it to fit the law. Trokhymovych and co-authors study that copy in a 2025 paper, asking how the fork changed what readers learn. Their method compares the two sites article by article at large scale.

The team compared more than 1.9 million articles across the two sites. They sorted the edits by place, time, topic, and text changes to spot patterns of control. From that work they built classes for the main kinds of knowledge shaping in the fork, with counts for how wide each kind runs. The method could also be used to study other wiki forks and group-edited sites.

Catalog record

CREATORMykola Trokhymovych, Oleksandr Kosovan, Nathan Forrester, Pablo Aragón, Diego Saez-Trumper, Ricardo Baeza-Yates
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE49 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2504-10663

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Characterizing Knowledge Manipulation in a Russian Wikipedia Fork Mykola Trokhymovych, Oleksandr Kosovan, Nathan Forrester, Pablo Aragón, Diego Saez-Trumper, Ricardo Baeza-Yates Abstract Wikipedia is powered by MediaWiki, a free and open-source software that is also the infrastructure for many other wiki-based online encyclopedias. These include the recently launched website Ruwiki, which has copied and modified the original Russian Wikipedia content to conform to Russian law. To identify practices and narratives that could be associated with different forms of knowledge manipulation, this article presents an in-depth analysis of this Russian Wikipedia fork. We propose a methodology to characterize the main changes with respect to the original version. The foundation of this study is a comprehensive comparative analysis of more than 1.9M articles from Russian Wikipedia and its fork. Using meta-information and geographical, temporal, categorical, and textual features, we explore the changes made by Ruwiki editors. Furthermore, we present a classification of the main topics of knowledge manipulation in this fork, including a numerical estimation of their scope. This research not only sheds light on significant changes within Ruwiki, but also provides a methodology that could be applied to analyze other Wikipedia forks and similar collaborative projects. 1 Universitat Pompeu Fabra 2 Ukrainian Catholic University 3 Independent Scholar {mykola.trokhymovych, pablo.aragon, diego.saez}@upf.edu, [email protected], [email protected], [email protected] 1 Introduction 0 0 footnotetext: To appear in ICWSM’25. Online information dissemination plays a key role in shaping public opinions and attitudes. As the the world’s largest encyclopedia and the ninth most visited website globally, 1 1 1 https://www.similarweb.com/website/wikipedia.org , Accessed 12 September 2024. Wikipedia holds an influential position within the web ecosystem ( Piccardi et al. 2021 ) . It is maintained through a collaborative community effort to become the “sum of all human knowledge” ( Sutcliffe 2016 ) . That is, anyone can freely edit Wikipedia content, although several policies and guidelines decided by the community must be followed to guarantee the quality and integrity of knowledge ( McDowell and Vetter 2020 ) . Given that its content empowers various applications, for example, integrating verified facts into curricula ( Lemmerich et al. 2019 ) , fact-checking ( Trokhymovych and Saez-Trumper 2021 ) , or training of large language models ( Devlin et al. 2019 ) , knowledge on Wikipedia has a major societal impact. There are actors who are not comfortable with the content and policies of Wikipedia. For example, some states like China and Turkey have repeatedly blocked access to the website ( Sezer and Dolan 2017 ; Siegel 2019 ) . In the case of China, the decision to censor Wikipedia was also followed by the launch of Baidu Baike , an online encyclopedia with content in accordance with the requirements of the Chinese Government ( Woo 2007 ; Siegel 2019 ) . Moreover, the open-source nature of MediaWiki , the software that powers Wikipedia, has enabled the launch of alternative wiki-based encyclopedias. A well-known example is Conservapedia , created by detractors of Wikipedia’s core policy of neutrality and self-described as American conservative and fundamentalist Christian ( Johnson 2007 ) . Another example is Runiversalis , the content of which must follow the requirements of the Russian legislation and its traditional values ( Runiversalis 2024 ) . While the views on these encyclopedias don’t compare to Wikipedia’s, the growing cultural prominence of alternative facts has sparked a noticeable rise in both traffic and interest ( Fitts 2017 ) . Figure 1: Summary of our research that analyzes changes in a Russian Wikipedia fork to assess knowledge manipulation. Russian Wikipedia appeared in May 2001 during the first wave of non-English Wikipedias. In June 2023, a fork of Russian Wikipedia was launched online, hereinafter referred to as RWFork . The project was founded by Vladimir Medeyko, former Director of Wikimedia Russia, a Wikimedia Chapter organization. While this project was powered by MediaWiki software like the aforementioned alt-Wikipedias, its content was also copied from Russian Wikipedia and later edited to conform to the Russian legislation ( Cohen 2023 ) . Therefore, RWFork is an organized effort to manipulate knowledge, originally created with neutral editorial policies, in order to comply with the editorial policies of a specific state. Research Questions. In this work, we examine a Russian Wikipedia fork. Our primary goal is to characterize how original content from Russian Wikipedia has been manipulated (see Figure 1 ). To achieve this, the initial step is to identify which articles have undergone changes and their relevance. Therefore, our first research question is: • RQ1 How relevant are Russian Wikipedia articles changed by RWFork editors? Next, we aim to examine changes made within RWFork in detail to characterize the editorial process and the nature of content modifications. This leads to our second research question: • RQ2 How has article content changed in RWFork ? Finally, we are interested in categorizing the changes that have occurred in RWFork to provide a clear understanding of the broader patterns of knowledge manipulation. As a result, our third research question is: • RQ3 What are the patterns of knowledge manipulation in RWFork ? Methodology. To answer our research questions, we have developed a methodology to compare the content of two MediaWiki-powered websites: Russian Wikipedia and RWFork . The first challenge we face is data collection and preparation. Relying on MediaWiki APIs, we implemented a data retrieval approach to extract data from both sources. To address RQ1 , we use bootstrapping to estimate the relevance metrics of articles along with their confidence intervals. Later we use these estimations to characterize the articles that were changed by RWFork editors. As for RQ2 , we use exploratory data analysis on a wide range of temporal, geographical, categorical, and topical features. Our methodology also includes named entity recognition to identify the most frequent entities in deletions and additions. To answer RQ3 , we combine advanced natural language processing tools, clustering algorithms, and qualitative analysis to build a classification of the main topics affected by knowledge manipulation. Main Findings. By applying our methodology to a dataset with more than 1.9M articles of the Russian Wikipedia and its fork, we discover that: • RQ1 : RWFork editors have modified articles that have a considerably higher number of page views compared to others, hence influential ones. We find that 1.75% of articles ( changed pages) generate 14.2% of the total page views. Also, according to our analysis, modified pages refer to controversial topics, and have almost twice the higher revert rate than duplicated ones. • RQ2 : Unlike Wikipedia editors, RWFork editors have a shorter activity period that aligns with standard office hours and reduced weekend activity. Most frequently altered pages pertain to Ukraine, Russia, and Belarus locations. The most frequently added and deleted categories are related to the 2022 Russian invasion of Ukraine, particularly occupied territories or sanctions. Also, we discover frequent additions of Russian Government resources, contrasting with the frequent deletion of EU and Ukrainian Government websites. • RQ3 : We find that most of the content changes ( \sim 90%) can be classified into 8 main categories. In particular, we discover that the majority of the content changes are knowledge manipulations related to territory reassignment, international sanctions, and terminology variations related to the 2022 Russian invasion of Ukraine (

Read the complete text