Reducing the information disparity by languages in online knowledge service

Abstract

During the COVID-19 pandemic, it has been important to analyze public attention to understand the potential harm of fake news and to direct people toward more reliable sources of information. The current study shares a collection of public Web resources on COVID-19 from Wikipedia, which is one of the most extensive information sources on the Internet. Tracking relevant content can be difficult because page titles may change over time and may not contain any relevant keywords. To address this issue, the research developed a new search method that identifies relevant content more comprehensively than keyword-based searches. This method allows an intuitive categorization of hierarchical topics such as biomedical, people-related, and regional events. A total of 18,492 Wikipedia pages on coronavirus disease 2019 (COVID-19) in 11 languages over 852 days from January 1, 2020, are shared via this research. By correlating the confirmed cases in various countries with the view counts of Wikipedia pages in corresponding languages, we were able to study how information-seeking behavior on Wikipedia changed as the pandemic evolved. This analysis revealed the prevalence of seeking information in Wikipedia with different immediacy depending on language, that as the pandemic progressed, public attention shifted from biomedical information to regional events. Furthermore, Wikipedia's voluntary nature leads to inevitable differences in the scale and depth of information available across the urgent matter among several hundreds of language projects it supports. To study this gap, we analyze a near-complete set of 7,830 articles on COVID-19 written in 15 languages and examine how various information types, particularly on \textit{local} content that is about a specific geographic region, are covered across the studied languages. We also examine the temporal relationship across language projects regarding information propagation. Overall, the findings suggest that there is an increasing need for local information during the pandemic and that additional support is needed to cover this type of content strategically.