liveBook · Manning

Entity linking

As explained in Build AI Drug Discovery Pipelines

Entity linking is the computational task of mapping a textual or database mention to the canonical entity it denotes.

Entity linking maps a textual or database mention to the canonical entity it denotes. It resolves inconsistent identifiers so data from different sources can be connected. section 13.2.4

In multimodal drug-discovery platforms, entity linking supports data harmonization by connecting literature, databases, and knowledge-graph records that refer to the same biological entity. section 13.2.4

Resolving a protein identifier

A paper mentioning “JAK2” can be linked to Entrez Gene ID 3717, allowing a knowledge graph to treat both references as the same protein kinase and answer queries spanning databases. section 13.2.4

Entity linking differs from ontology mapping: linking identifies the same entity across references, while ontology mapping aligns different terminology systems describing related concepts. section 13.2.4

How Build AI Drug Discovery Pipelines covers it

How deep it goes
Defines the mapping task and demonstrates its integration role, but stops short of specifying linking algorithms or evaluation metrics.
Examples
Uses the JAK2 mention and Entrez Gene ID 3717 as a cross-database protein-resolution example.
What you can do after
Apply it to connect inconsistent biological references and retrieve knowledge across databases.
How it connects
Connects canonical identifiers with data harmonization, knowledge graphs, multimodal representations, and ontology mapping.

Also explained in

Compare how 2 books explain entity linking →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Build AI Drug Discovery Pipelines. The section links point to the parts of the book it draws on.