NILK: Entity Linking Dataset Targeting NIL-linking Cases

Anastasiia Iurshina, Jiaxin Pan, Rafika Boutalbi, Steffen Staab

Abstract: The NIL-linking task in Entity Linking deals with cases where the text mentions do not have a corresponding entity in the associated knowledge base. NIL-linking has two sub-tasks: NIL-detection and NIL-disambiguation. NIL-detection identifies NIL-mentions in the text. Then, NIL-disambiguation determines if some NIL-mentions refer to the same out-of-knowledge base entity. Although multiple existing datasets can be adapted for NIL-detection, none of them address the problem of NIL-disambiguation. This paper presents NILK, a new dataset for NIL-linking processing, constructed from WikiData and Wikipedia dumps from two different timestamps. The NILK dataset has two main features: 1) It marks NIL-mentions for NIL-detection by extracting mentions which belong to newly added entities in Wikipedia text. 2) It provides an entity label for NIL-disambiguation by marking NIL-mentions with WikiData IDs from the newer dump. We make available the annotated dataset along with the code1. The NILK dataset is available at: https://zenodo.org/record/6607514