Structuring wiki revision history

Abstract

Revision history of a wiki page is traditionally maintained as a linear chronological sequence. We propose to represent revision history as a tree of versions. Every edge in the tree is given a weight, called adoption coefficient, indicating similarity between the two corresponding page versions. The same coefficients are used to build the tree. In the implementation described, adoption coefficients are derived from comparing texts of the versions, similarly to computing edit distance. The tree structure reflects actual evolution of page content, revealing reverts, vandalism, and edit wars, which is demonstrated on Wikipedia examples. The tree representation is useful for both human editors and automated algorithms, including trust and reputation schemes for wiki.