Entity Linking for low-resource African languages faces a major bottleneck: orthographic variation.
This gold standard validation dataset contains:
356 sentences (filtered from Wolof-MasakhaNER 2.0 validation split to exclude sentences without entity mentions or with only DATE mentions).
695 entity mentions categorized into LOC, ORG, and PER.
584 linkable mentions (In-KB) resolved to 276 unique Wikidata QIDs.