Identifying Sock-Puppets on Wikipedia

The research tested a semantic clustering technique to identify coordinated manipulation on Wikipedia, focusing on the Russo-Ukrainian war and 48 related pages. While the method did not consistently distinguish between legitimate and illegitimate editors globally, it successfully identified several dense clusters of sock-puppet accounts and coordinated IP-based editing. The study concludes that semantic clustering is a useful tool for grouping accounts with highly similar editing patterns, which can then be targeted for human-led investigation.

• The semantic clustering method identified 176 clusters of editors, but it did not consistently separate legitimate editors from those engaging in illegitimate activity, as clusters often reflected shared themes and topics rather than policy violations.
• The analysis identified six specific clusters of high interest containing dense groups of sock-puppet accounts or coordinated IP-range activity, where multiple accounts were linked to a single user.
• Cluster 1 contained ten editors, including five sock-puppets (three linked to one user), who focused on changing the description of Ukraine's location from Eastern to Central Europe in the 'Ukraine' article.
• Cluster 12 consisted of eight editors, including four sock-puppets linked to a single user, who engaged in an edit war over casualty details and personal information in the 'Casualties of the Russo-Ukrainian War' article.
• Cluster 78 included ten editors, eight of whom were blocked for reasons including vandalism and sock-puppetry, primarily fighting an edit war over whether to name the country 'Ukraine' or 'The Ukraine'.
• Cluster 100 comprised nine editors, including four sock-puppets (two linked to one user), who focused on updating country statistics and military data in the 'Russia–Ukraine relations' and 'Armed Forces of Ukraine' articles.
• Cluster 0 consisted of seven IP accounts from the same range, all blocked, which were used to repeatedly remove links to the 'Malaysia Airlines Flight 17' article from the 'Russia–Ukraine relations' page.
• Cluster 25 contained five editors, four of whom were IP accounts from a range blocked for disruptive editing and adding links to antisemitic and conspiracist content, focusing on Switzerland and Swiss banking.
• The study highlights that automated analysis of Wikipedia is hindered by the unstructured nature of 'diffs' and the need for manual navigation of user pages and investigation archives to link sock-puppets to principal accounts.

APA

Chavana, N. P., Inskip, C., Miller, C., & Weir, D. (n.d.). Identifying Sock-Puppets on Wikipedia. Institute for Strategic Dialogue. https://www.isdglobal.org/wp-content/uploads/2024/04/Identifying-Sock-Puppets-on-Wikipedia_A-Semantic-Clustering-Approach.pdf

Chicago

Chavana, Nestor Prieto, Chris Inskip, Carl Miller, and David Weir. Identifying Sock-Puppets on Wikipedia. Institute for Strategic Dialogue, n.d. https://www.isdglobal.org/wp-content/uploads/2024/04/Identifying-Sock-Puppets-on-Wikipedia_A-Semantic-Clustering-Approach.pdf.

Wikipedia

{{cite report |last1=Chavana |first1=Nestor Prieto |last2=Inskip |first2=Chris |last3=Miller |first3=Carl |last4=Weir |first4=David |title=Identifying Sock-Puppets on Wikipedia |publisher=Institute for Strategic Dialogue |url=https://www.isdglobal.org/wp-content/uploads/2024/04/Identifying-Sock-Puppets-on-Wikipedia_A-Semantic-Clustering-Approach.pdf |access-date=23 October 2026 |via=Climate Insights Directory}}

BibTeX

@techreport{chavanandidentifying, author = {Chavana, Nestor Prieto and Inskip, Chris and Miller, Carl and Weir, David}, title = {{Identifying Sock-Puppets on Wikipedia}}, institution = {Institute for Strategic Dialogue}, url = {https://www.isdglobal.org/wp-content/uploads/2024/04/Identifying-Sock-Puppets-on-Wikipedia_A-Semantic-Clustering-Approach.pdf}, urldate = {2026-10-23}, note = {Indexed by Climate Insights Directory} }