Title: Collaboratively fixing automatic citations in Wikipedia with Web2Cit
Speaker: Diego de la Hera (User:Diegodlh)
Abstract: Automatic citations in Wikipedia sometimes fail to retrieve accurate metadata, which may particularly affect non-mainstream sources. Web2Cit is a tool that allows technical and non-technical volunteers alike to participate in improving automatic citations. In this hands-on workshop you will learn how you can engage with the Web2Cit community to collaboratively fix automatic citations for sources relevant to your communities.

Wikipedia’s automatic citation generator relies on source webpages’ correct embedding of citation metadata. When this is not the case, which often happens, individual algorithms tailored to specific websites must be programmed. However, some websites take longer to be addressed and algorithms tend to break when websites change. This results in wrong citation metadata retrieved from some of them, which particularly affects non-mainstream websites.

Web2Cit is a tool recently released to help fix automatic citations, requiring much less technical skills than were needed before. In addition, it is collaborative, meaning that community members may contribute according to their skills, enabling a wider diversity of contributors to participate.

Web2Cit documentation and training material are available online. But, as with any tool, starting to use it can be challenging. For this reason, a few demonstrations have been organized in the past. However, most of them were not effectively hands-on, and all of them occurred in the middle of the night in the ESEAP region.

In this hands-on workshop we will particularly focus on one specific and particularly annoying automatic citation error: misrecognition of the source type (e.g., “webpage” instead of “newspaper article”). Specifically, we will learn how to: (1) use the Web2Cit user script to use collaborative automatic citations in addition to regular automatic citations; (2) configure extraction tests to indicate the expected output for a given webpage; (3) configure a simple extraction procedure to match the extraction test; and (4) keep track of test results to fix procedures in case they break.

If possible, bring the Web2Cit user script installed: https://meta.wikimedia.org/wiki/Web2Cit#Install . You may also bring a URL you would like to work with; otherwise, you may choose one from a list to be provided during the session.
Presented at: Suntec Singapore Convention and Exhibition Centre, 2023-08-15 06:15 to 07:15 UTC