Automatic Verification of References of Wikidata Statements

Elena Simperl, Odinaldo Rodrigues, Albert Meroño-Peñuela, Kholoud Saad alghamdi, Gabriel Maia Rocha Amaral, Jongmo kim, Miriam Redi, Bohui Zhang, Yihang Zhao

Confirmation: I have read and agree with the workshop's policy on behalf of myself and my co-authors.

Keywords: reference verification, quality assurance, tools, Wikidata

TL;DR: Introducing ProVe: a tool to automatically verify and assess the quality of the references in Wikidata

Abstract: Wikidata is one of the world's most important machine-readable data assets. It is used by web search engines, virtual assistants such as Siri and Alexa, fact checkers, and in over 800 projects in the Wikimedia ecosystem. Wikidata contains information about 100 million of topics edited daily by 24 thousand active editors. Manually checking whether an individual reference supports the claim of a Wikidata statement is not very difficult, but it is a slow and somewhat tedious process. Given the overall number of statements to check, the collaborative nature of the knowledge graph, and the fact that referenced documents can change over time, preserving the quality of the references is an onerous process requiring continuous intervention. In this paper, we present ProVe - a reference verification tool for Wikidata statements developed from state-of-the-art research for quality assurance in collaboratively constructed knowledge graphs. ProVe harnesses the power of LLMs to automatically verify and assess the quality of the references in Wikidata.

Format: Poster

Submission Number: 5