WikidataComplete – An Easy-to-Use Method for Rapid Validation of Text-Extracted New Facts Applied to the Wikidata Knowledge Graph

Abstract

Wikidata is one of the most used knowledge graphs (KG) and it plays a vital role in the Semantic Web community. Many industries have integrated Wikidata into solutions dedicated to intelligent assistance, information retrieval, or knowledge integration. As one of the biggest KGs, Wikidata receives millions of edits every year. However, it is still far from complete. Generally, a natural workflow to ingest a new fact into Wikidata starts by searching the relative information in a free-text document collection (e.g., Wikipedia articles). This information is used to create a new fact (or update a fact) on Wikidata. The entire process is labor-intensive. In this paper, we present WikidataComplete, a plugin that facilitates Wikidata editors to contribute to the completion of the Wikidata KG. For the implementation of WikidataComplete, we integrated the latest question-answering (QA) technologies in order to extract the new facts. We embed our fact-ingestion workflow directly on the Wikidata entity page to make the insertion of facts smooth and efficient. Ultimately, WikidataComplete can be a handy tool for Wikidata contributors, and it has the potential to complete millions of missing facts in the Wikidata KG.