Given a text with entity links, the task of entity aspect linking is to identify which aspect of an entity is referred to in the context. For example, if a text passage mentions the entity "USA'', is USA mentioned in the context of the 2008 financial crisis, American cuisine, or else? Complementing efforts of Nanni et al (2018), we provide a large-scale test collection which is derived from Wikipedia hyperlinks in a dump from 01/01/2020. Furthermore, we offer strong baselines with results and broken-out feature sets to stimulate more research in this area. Data, code, feature sets, runfiles and results are released under a CC-SA license and offered on our aspect linking resource web page http://www.cs.unh.edu/∼dietz/eal-dataset-2020/
CITATION STYLE
Ramsdell, J., & Dietz, L. (2020). A Large Test Collection for Entity Aspect Linking. In International Conference on Information and Knowledge Management, Proceedings (pp. 3109–3116). Association for Computing Machinery. https://doi.org/10.1145/3340531.3412875
Mendeley helps you to discover research relevant for your work.