📝 SummarXiv

Abstract

Wikipedia is among the largest examples of collective intelligence on the Web with over 61 million articles covering over 320 languages. Although edited and maintained by an active workforce of human volunteers, Wikipedia is highly reliant on automated bots to fill gaps in its human workforce. As well as administrative and governance tasks, these bots also play a role in generating content, although to date such agents represent the smallest proportion of bots. While there has been considerable analysis of bots and their activity in Wikipedia, such work captures only automated agents that have been actively deployed to Wikipedia and fails to capture the methods that have been proposed to generate Wikipedia content in the wider literature. In this paper, we conduct a systematic literature review to explore how researchers have operationalised and evaluated automated content-generation agents for Wikipedia. We identify the scope of these generation methods, the techniques and models used, the source content used for generation and the evaluation methodologies which support generation processes. We also explore implications of our findings to CSCW, User Generated Content and Wikipedia, as well as research directions for future development. To the best of our knowledge, we are among the first to review the potential contributions of this understudied form of AI support for the Wikipedia community beyond the implementation of bots.