Relating Wikipedia article quality to edit behavior and link structure

Thorsten Ruprechter*, Tiago Santos, Denis Helic

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review


Currently, the relation between edit behavior, link structure, and article quality is not well-understood in our community, notwithstanding that this relationship may facilitate editing processes and content quality on Wikipedia. To shed light on this complex relation, we classify article edits and perform an in-depth analysis of editing sequences for 4941 articles. Additionally, we build a network of internal Wikipedia hyperlinks between articles. Using this data, we compute parsimonious metrics to quantify editing and linking behavior. Our analysis unveils that conflicted articles differ substantially from others in almost all metrics, while we also detect slight trends for high-quality articles. With our network analysis we find evidence indicating that controversial and edit war articles frequently span structural holes in the Wikipedia network. Finally, in a prediction experiment we demonstrate the usefulness of edit behavior patterns and network properties in predicting conflict and article quality. With our work, we assist online collaboration communities, especially Wikipedia, in long-term improvement of content quality by offering valuable insights about the interplay of article quality, controversies and edit wars, editing behavior, and network properties via sequence-based edit and network-based article metrics.

Original languageEnglish
Article number61
JournalApplied Network Science
Issue number1
Publication statusPublished - 1 Dec 2020


  • Article quality
  • Conflict
  • Controversy
  • Edit behavior
  • Edit wars
  • Link structure
  • Semantic edit types
  • Wikipedia

ASJC Scopus subject areas

  • General
  • Computer Networks and Communications
  • Computational Mathematics


Dive into the research topics of 'Relating Wikipedia article quality to edit behavior and link structure'. Together they form a unique fingerprint.

Cite this