Tasks

Overview

Task 1: Named Entity Recognition and Classification (NERC)

• Subtask 1.1 - NERC Coarse-grained: this task includes the recognition and classification of entity mentions according to coarse-grained types (Person, Location, Organisation and Product).
• Subtask 1.2 - NERC Fine-grained: this task includes the recognition and classification of entity mentions according to fine-grained types (cf. column 2 in Table 2), plus the detection and classification of nested entities of depth 1, as well as entity mention components.

Task 2 : Named Entity Linking (EL)

This task includes the linking of named entity mentions to a unique referent in a knowledge base (KB) or to a NIL node if the mention does not have a referent in the KB. The chosen KB is Wikidata.

The entity linking task includes two settings: with and without prior knowledge of mention boundaries. Concretely speaking, the evaluation period will consist of two consecutive rounds, where a first NEL task without prior information on mentions will be evaluated during round 1 (i.e. task bundles 1 and 2), and a second one with information on mention boundaries (but no type) during the second round (bundle 5).

Detailed description

For a detailed description of the tasks and instructions relative to participation, download the HIPE - Shared Task Participation Guidelines.

Task bundles