Semi-automatic Assessment of Multiple Viewpoint Representation in Wikidata

This chapter delves into the representation of multiple viewpoints in Wikidata, focusing on controversial topics and the mechanisms used to identify and analyze them. The study employs a semi-automatic methodology to explore six key mechanisms: controversial terms in item descriptions, multiple values in single-value properties, properties indicating disputes, statement rankings, qualifiers showing ambiguity, and sentiment and clustering analysis of multilingual item descriptions. The research reveals that while Wikidata has the structural capacity to accommodate diverse perspectives, these mechanisms are underutilized. The analysis of item descriptions shows that controversial terms appear in fewer than 0.01% of cases, and properties explicitly denoting controversy are rare. However, the study highlights the potential of multilingual sentiment analysis and clustering to uncover differing viewpoints. The results indicate that items flagged as controversial typically belong to a narrow set of categories, and the use of statement rankings to surface controversial information is limited. The chapter concludes by suggesting that while Wikidata's mechanisms for detecting controversial items show promise, there is a need for further investigation and improvement in representing diverse viewpoints.