Entity identification finds the thing behind the name.
A word, name or phrase is not automatically a unique entity. Identification requires enough evidence to determine which person, organization, product, place, concept or other referent a mention is intended to denote.
The objective is not merely to detect an entity-like mention. It is to resolve that mention to the best-supported identity using type, attributes, aliases, relationships, context and applicable identifiers.
Detection finds a mention. Identification resolves its identity.
On this page, entity identification means determining which specific referent a textual mention denotes. In NLP practice, mention detection, named-entity recognition, entity linking and entity disambiguation are related but distinct tasks. Identity resolution becomes important when names are ambiguous, aliases differ, entity types overlap or the same entity appears in several linguistic forms.
Mention Detection
Locate a span of text that may denote an entity, without yet claiming which specific entity it is.
WHERE IS THE MENTION?Entity Identification
Resolve the mention to the best-supported referent using context, type, attributes, relationships and available identifiers.
WHICH REFERENT IS INTENDED?Entity Salience
Estimate how central an already identified entity is to the meaning of a specific document or passage.
HOW CENTRAL IS IT?Entity Authority
Analyze broader recognition, reputation, corroboration and source context around an already identified entity.
HOW WELL IS IT ESTABLISHED?Same surface form. Different underlying entities.
Select an ambiguous mention. These are illustrative context windows showing how type, neighboring language, attributes and relations can change the best-supported entity resolution.
The surrounding product and hardware terms strongly support the technology-company interpretation rather than the common-noun fruit interpretation.
Type narrows the candidate space.
Type can sharply narrow the candidate space. The categories below are a practical editorial taxonomy for this page, not a reproduction of one vendor’s entity-type enum. Different NLP systems use different type inventories.
Person
Names, roles, biographies, affiliations, dates and works can distinguish individuals with similar names.
Organization
Legal names, domains, products, addresses, identifiers and affiliations help resolve companies and institutions.
Location
Country, region, coordinates, landmarks and containment relationships disambiguate places with shared names.
Product
Manufacturer, model, category, version and product-family relations distinguish individual goods or services.
Creative Work
Creator, release date, medium, edition and title context separate books, films, songs and other works.
Concept
Definitions, disciplines, mechanisms and adjacent terminology distinguish abstract topics and technical ideas.
Event
Date, location, participants and event series can turn a generic name into a unique occurrence.
Typed Value
Some NLP systems also classify addresses, dates, numbers, prices and similar structured values even though they are not conventional named entities.
One entity can appear under many surface forms.
Aliases and coreference are related identity problems, but they are not identical. Alternate names and acronyms can map directly to one entity, while pronouns or phrases such as “the company” require discourse-level coreference resolution before they can be attached to that identity.
Meaning comes from neighboring evidence.
The same mention can resolve differently depending on the surrounding words. Local context may provide profession, product, location, action, category or relationship signals that eliminate incompatible candidates.
Identity emerges from matching attributes.
A candidate becomes stronger when the attributes in the document agree with attributes associated with that entity. No single attribute needs to be decisive; the fingerprint is created by the pattern.
Relations often resolve what names alone cannot.
Connections to organizations, products, places, roles, dates and other entities form a graph fingerprint. When several relations converge on one candidate, identity becomes much easier to resolve.
Explicit identifiers reduce avoidable ambiguity.
Structured data does not replace readable page content. For Organization markup, Google currently documents fields such as name, alternateName, URL, sameAs, legalName, taxID, vatID and iso6523Code, and states that some organization properties can help disambiguate an organization in Search. Use only fields that actually apply to the entity.
{ "@context": "https://schema.org", "@type": "Organization", "name": "Example Corporation", "alternateName": "Example Corp", "url": "https://example.com", "sameAs": [ "https://social.example/example" ], "legalName": "Example Corporation Ltd.", "iso6523Code": "0199:<VALID_LEI>" }
Consistent naming reduces unnecessary identity fragmentation across the site and external profiles.
Alternate naming helps connect common abbreviations or public-facing names to the same organization.
Official and external profile URLs can provide useful identity context when they accurately describe the same entity.
For organizations, Google documents fields such as ISO 6523 identifiers, legal name, tax ID and VAT ID when relevant.
Identification weakens when identity evidence conflicts.
This section stays at the identification layer: it flags identity-evidence problems without replacing the deeper candidate-resolution methods covered in ENTITY / 06 — Entity Disambiguation.
Identification should survive every page layer.
A well-identified entity remains coherent from the title and opening definition through attributes, examples, relationships, structured identity and contextual internal links.
Correct synthesis begins with correct identity.
Retrieval and answer synthesis can select the wrong evidence when an ambiguous mention is mapped to the wrong referent. Entity resolution is therefore an information-quality dependency for systems that retrieve or synthesize entity-specific evidence.
Audit whether a machine can tell which entity you mean.
The audit tests identity clarity rather than keyword frequency. A good result means the page provides enough consistent evidence to distinguish the intended entity from plausible alternatives.
Not every signal carries the same resolving power.
Entity identification becomes more defensible when several compatible evidence layers converge. The order below is an internal review model, not a universal ranking of evidence strength and not a public search-engine scoring system.
The surface form creates the initial candidate set but may remain highly ambiguous.
CANDIDATE GENERATIONPerson, organization, product, place or concept classification can eliminate entire candidate classes.
CLASS NARROWINGNearby terms, actions and domain language reveal which interpretation fits the passage.
CONTEXT FITDistinctive properties such as profession, location, product family or role strengthen one identity over alternatives.
FINGERPRINTAffiliations, products, people, places and other defensible connections can collapse remaining ambiguity.
GRAPH FITApplicable official URLs, IDs or explicit structured identity fields can provide strong anchoring when accurate.
EXPLICIT ANCHORResolution improves when incompatible candidates fail the evidence test.
This conceptual matrix uses “Jordan” in a basketball passage to show why candidate elimination can be more useful than simply accumulating mentions.
Move from ambiguity to well-supported identity.
These states describe an internal review workflow. They are not Google confidence levels or universal machine thresholds.
The mention maps naturally to several plausible entities and context is insufficient.
DO NOT FORCE RESOLUTIONType or domain context removes some candidates, but multiple reasonable interpretations remain.
MORE EVIDENCE NEEDEDMost signals support one candidate, but a material alternative has not been fully eliminated.
QUALIFY IF NECESSARYType, context, attributes and relationships consistently converge on one referent.
LOW AMBIGUITYThe representation also includes applicable authoritative identifiers or canonical destinations aligned with the same entity.
IDENTITY ANCHOR PRESENTGround the model in documented identity systems.
These primary sources support the factual boundaries used on this page. Google Cloud documents entity analysis, types, mentions, metadata and salience; Google Search documents Organization structured data and says some organization properties can help disambiguate an organization. None of these documents establishes a public Google Search “Entity Identification” ranking factor.
Natural Language API Basics
Documents entity analysis, entity types, metadata and mentions, including the use of type information to distinguish similarly named entities.
PRIMARY SOURCE → GOOGLE CLOUD / ENTITY ANALYSISAnalyzing Entities
Shows detected entity names, types, mentions and metadata; current examples include Knowledge Graph MIDs or Wikipedia URLs for some entities when available.
PRIMARY SOURCE → GOOGLE SEARCH / STRUCTURED IDENTITYOrganization Structured Data
Documents Organization properties including name, alternateName, URL, sameAs, legalName, taxID, vatID and iso6523Code, and notes that some properties are used to help disambiguate organizations.
PRIMARY SOURCE →Research the entity layer. Explore the systems behind identity and meaning.
Explore the Entity SEO research layer from identification and attributes through relationships, disambiguation, salience, contextual relevance, coverage, representation depth, authority and AI Search. Each node owns a distinct analytical function.
What Is Entity SEO?
Define the entity-oriented layer through identity, attributes, relationships, context and disambiguation.
OPEN NODE → ENTITY / 02 · DISTINCTIONEntities vs Keywords
Separate lexical strings from identifiable referents while preserving the role of keyword and query analysis.
OPEN NODE →Entity Identification
Resolve mentions to the best-supported referent through type, attributes, context, aliases, relationships and applicable identifiers.
CURRENT NODEEntity Attributes
Represent material properties that define, distinguish and enrich an entity without turning trivia into false depth.
OPEN NODE → ENTITY / 05 · GRAPHEntity Relationships
Model defensible connections between entities while keeping co-occurrence separate from verified relationship evidence.
OPEN NODE → ENTITY / 06 · RESOLUTIONEntity Disambiguation
Resolve competing candidates when names, aliases or semantic neighborhoods overlap.
OPEN NODE → ENTITY / 07 · IMPORTANCEEntity Salience
Analyze which identified entities are central, supporting or peripheral to a document or passage.
OPEN NODE → ENTITY / 08 · CONTEXTContextual Relevance
Evaluate how task, domain, surrounding concepts and local evidence affect an entity’s relevance in context.
OPEN NODE → ENTITY / 09 · COVERAGEEntity Coverage
Evaluate whether the information architecture represents the important entities and semantic neighborhoods required by the topic.
OPEN NODE → ENTITY / 10 · REPRESENTATIONEntity Representation Depth
Evaluate how deeply an entity is represented through attributes, relationships, context, distinctions and supporting information.
OPEN NODE → ENTITY / 11 · AUTHORITYEntity Authority
Examine broader recognition, reputation, corroboration and source context without inventing a public Google entity-authority score.
OPEN NODE → ENTITY / 12 · FRONTIEREntity SEO & AI Search
Explore how resolved identity and entity context can support retrieval and source interpretation without guaranteeing selection or citation.
OPEN FRONTIER →A name points toward a candidate. Context resolves the entity.
Entity identification becomes more defensible when naming, type, attributes, aliases, relationships and applicable identifiers converge on the same referent. The goal is not to mention more entities. It is to make the intended identity difficult to confuse.