Entity linking

As explained in Graph Algorithms for Data Science

Entity linking is the task of connecting a named-entity mention in text to the corresponding entity identifier in a knowledge base.

Entity linking connects a named-entity mention in text to the corresponding identifier in a knowledge base. It resolves ambiguity such as distinguishing two people with the same name and enables external data enrichment. section 12.2.1 section 12.6

A practical linking workflow typically involves:

• Identify entities using named entity recognition or an information extraction pipeline. section 12.4.4 section 12.6
• Search a target knowledge base such as WikiData, Wikipedia, DBpedia, or a biomedical resource. section 12.2.1 section 12.6
• Select the correct identifier while handling aliases, abbreviations, ambiguous names, missing entries, and insufficient context. section 12.2.1 section 12.4.4

A basic WikiData lookup

A Python function can call the WikiData API, search an extracted entity, and return the first result's URL and description. For example, Steve Jobs maps to Q19837, while an entity without a WikiData entry returns None. Choosing the first result is a useful initial implementation, but ignoring sentence context can incorrectly link an ambiguous name or phrase. section 12.4.4

Figure 12.5 Named entity linking

Use the named-entity linking figure to follow how mentions become links to knowledge-base entities. section 12.2.1

How Graph Algorithms for Data Science covers it

How deep it goes

Explains the task, challenges, and a first-result WikiData implementation, but defers advanced context-based disambiguation.

Examples

Uses Steve Jobs, CPU, ambiguous Eric Schmidt, unavailable AlgoTrim, and contextually mislinked phrases as examples.

What it assumes

Assumes extracted named entities for linking-only models while explaining entity linking, disambiguation, and knowledge-base targets directly.

What you can do after

Apply it via the WikiData API to map extracted entities to identifiers and inspect descriptions for basic validation.

How it connects

Connects named entity recognition, information extraction, knowledge bases, relation extraction, and graph representation.

liveBookmanning.com

Read Graph Algorithms for Data Science on liveBook Get the book at manning.com

Related concepts

• knowledge baseuses · A knowledge base is an organized collection of information that can be accessed or queried to support reference, retrieval, or reasoning.
• structured extractionpart of · Structured extraction is the process of identifying specified information in source content and representing it as organized data in a defined structure.
• disambiguationaddresses · Disambiguation is the process of resolving ambiguity by selecting the intended interpretation, entity, or goal among plausible alternatives.

This explainer was generated by Manning's knowledge graph from the text of Graph Algorithms for Data Science. The section links point to the parts of the book it draws on.