Entity Linking

The task of identifying mentions of entities in text and linking them to entries in a knowledge base. Bridges unstructured text and structured knowledge.

The Challenge

Text mentions are ambiguous:

• “Apple” → Company? Fruit?
• “Washington” → George? State? DC?
• “Michael Jordan” → Basketball player? Professor?

Process

1. Mention Detection: Find spans that refer to entities
2. Candidate Generation: Retrieve possible KB entries
3. Disambiguation: Select correct entry based on context
4. NIL Handling: Manage mentions with no KB entry

Methods

Traditional: String matching + feature engineering Neural: Learned embeddings for mentions and entities Hybrid: Combine string features with semantic similarity

Why It Matters

Entity linking enables:

• Knowledge graph population from text
• Question answering with KB grounding
• Document enrichment with structured metadata
• Cross-document coreference

Related: 06-molecule—knowledge-graph-construction, 06-molecule—ontology-design-patterns