Improving Wikipedia quality across languages

EDIC candidacy exam
Exam president: Prof. Karl Aberer
Thesis advisor: Prof. Robert  West
Co-examiner: Prof. Pierre Dillenbourg

Abstract
Since its creation in 2001, Wikipedia has become the reference platform for the free knowledge sharing. In these years, thanks to a very active community of volunteers, the website extended its coverage to 291 languages and many million topics. Despite this impressive success, the current versions of the encyclopedia present many limitations. For example, in the English version, the editors flagged 37% of the articles as incomplete or with poor quality. Furthermore, the knowledge is not uniform across all the languages and, although the English Wikipedia has several million articles, a large portion of the available languages have less than 100k pages.
In this report, I will give an overview of how I plan to address this issue, and of the related work on which I will build upon. The proposed solution is a recommender system that supports the editors in the writing of one article by providing suggestions about missing content. This system will be a way to reduce the onboarding barrier and to guide the newcomers in preparing high quality and consistent content. The recommendations will come from two main sources: inter-language, by exploiting content from other Wikipedia editions, and intra-language, by extracting information from articles with similar type.

Background papers
Omnipedia: bridging the wikipedia language gap, by  Patti Bao et al. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (CHI '12). 
MultiWiBi: The multilingual Wikipedia bitaxonomy project, by Tiziano Flati et al.  Artificial Intelligence 241 (2016).
Growing Wikipedia Across Languages via Recommendation, by  E. Wulczyn et al.  In Proceedings of the 25th International Conference on World Wide Web (WWW '16).