Entity Linking (EL) can be defined in two approaches:
- first approach (End-to-End): processing a piece of text to extract the entities (i.e. Named Entity Recognition) and then link these extracted entities to their counterpart in a given knowledge base (i.e. Wikipedia).
- second approach: contrary to the first approach, this one directly takes the annotated entities as input and then has just to link them against their counterpart in a given knowledge base (i.e. Wikipedia).
More in details in this survey
AIDA CoNLL-YAGO Dataset
The AIDA CoNLL-YAGO Dataset contains assignments of entities to the mentions of named entities annotated for the original CoNLL 2003 NER task. The entities are identified by YAGO2 entity name, by Wikipedia URL, or by Freebase mid. Approaches are evaluated based on span-based F1.
|Approach||F1||Paper / Source|
|Radhakrishnan et al. (2018)||93.7||ELDEN: Improved Entity Linking using Densified Knowledge Graphs|
|Le et al. (2018)||93.07||Improving Entity Linking by Modeling Latent Relations between Mentions|