Question Answering for Link Prediction and Verification

Maria Khvalchik, Artem Revenko, Christian Blaschke

Abstract: In this work we tackle the link prediction task in knowledge graphs. Following recent success of Question Answering systems in outperforming humans, we employ the developed tools to identify and verify new links. To identify the gaps in a knowledge graph, we use the existing techniques and combine them with Question Answering tools to extract concealed knowledge. We outline the overall procedure and discuss preliminary results.