A content-driven reputation system for the wikipedia

摘要: We present a content-driven reputation system for Wikipedia authors. In our system, authors gain when the edits they perform to articles are preserved by subsequent authors, and lose their rolled back or undone in short order. Thus, author is computed solely on basis of content evolution; user-to-user comments ratings not used. The we compute could be used flag new contributions from low-reputation it allow only with high contribute controversialor critical pages. A also provide an incentive high-quality contributions. have implemented proposed analyze entire Italian French Wikipedias, consisting total 691, 551 pages 5, 587, 523 revisions. Our results show that notion has good predictive value: changes performed significantly larger than average probability having poor quality, as judged human observers, being later undone, measured algorithms.