A package for loading data from wikilinks files. The package is designed for performing the following tasks:
-
Finding ambiguous entities.
Parameters:
- number of required mentions for each annotation;
- number of required annotations.
-
Finding different annotations for a given surface form.
-
Finding synonyms for a given annotation.
Annotation is always a link to Wikipedia URL. Redirections are taken into account, i.e. the output annotation is the final one after all redirections.
To learn more about wikilinks dataset see the technical report or visit their homepage.