Event Analysis Through QuoteKG: A Multilingual Knowledge Graph of Quotes

AbstractSignificant moments in history are often remarked upon by public figures in the form of quotes. As evidence of character traits and future political or personal decisions, quotes provide insight into the actions of their originators. The impact of a quote crosses language barriers and influences the public’s reaction to specific political stances. Nevertheless, effectively collating, attributing and analysing these quotes across languages remain challenging. Existing efforts have made strides in quote collections and analyses, yet several limitations persist, including a lack of context information, a labour-intensive extraction process and missing alignment of quote mentions across languages. Building upon QuoteKG, a multilingual knowledge graph of quotes that already addresses some of the aforementioned limitations, we present an approach for aligning quotes with event knowledge. QuoteKG is based on Wikiquote, a free and collaboratively created collection of quotes in many languages. Containing nearly one million quotes in 55 languages said by 69,000 people of public interest, QuoteKG extracts and aligns different mentions and contexts of quotes across a wide range of topics. We show that QuoteKG can be aligned with event knowledge. We use this alignment to enrich and analyse event-centric information by providing rich semantic context to important world events. QuoteKG is publicly available and can be accessed via a SPARQL endpoint.