Schema.org Type

Schema.org types are organized in a multi-level hierarchy where subtypes inherit properties from their parent types. This allows for both broad and highly specific entity definitions.

• Thing is the root type, from which all others descend
• CreativeWork inherits from Thing and adds properties like author and datePublished
• Article inherits from CreativeWork and adds articleBody and wordCount
• NewsArticle inherits from Article and adds dateline and printEdition

This inheritance chain means a NewsArticle automatically supports all properties from Thing, CreativeWork, and Article without explicit redefinition.

Each Schema.org type specifies expected properties with defined value types—either primitive data types like Text, Number, and Date, or references to other Schema.org types.

• A Product type expects offers to reference an Offer type
• A Person type expects address to reference a PostalAddress type
• A Recipe type expects recipeIngredient as Text and cookTime as Duration

This strict typing enables machines to parse and validate the data structure, ensuring that an AI system can reliably traverse the entity graph without encountering unexpected data shapes.

Certain properties accept only predefined enumeration values rather than free-form text, ensuring semantic consistency across implementations.

• Event status must be one of: EventScheduled, EventPostponed, EventRescheduled, EventCancelled
• Offer availability must be: InStock, OutOfStock, PreOrder, Discontinued
• MedicalEntity legal status uses: Approved, Withdrawn, Suspended

Using these controlled vocabularies eliminates ambiguity. An AI model interpreting EventCancelled knows definitively that the event will not occur, whereas a free-text field reading "canceled" or "cancelled" introduces parsing variability.

Schema.org allows an entity to be declared as multiple simultaneous types using JSON-LD's @type array syntax. This enables precise modeling of entities that serve multiple roles.

• A local business can be both a Restaurant and a BarOrPub
• A blog post can be both an Article and a HowTo
• A product page can be both a Product and a SoftwareApplication

This multi-typing capability avoids the limitations of rigid single-inheritance taxonomies and allows AI systems to understand the full scope of what an entity represents without forcing artificial categorization choices.

Beyond the core vocabulary, Schema.org includes hosted and external extensions for specialized industries, allowing domain-specific precision without bloating the base specification.

• pending.schema.org hosts proposed types under review, such as EducationalOccupationalCredential
• bib.schema.org extends types for bibliographic data like ComicStory and Thesis
• auto.schema.org provides automotive-specific types like MotorizedBicycle and BusOrCoach
• health-lifesci.schema.org adds medical types including Drug, MedicalCondition, and ImagingTest

These extensions allow enterprises in niche verticals to achieve granular semantic markup while the core vocabulary remains broadly applicable.

Schema.org maintains backward compatibility by marking outdated types as superseded rather than removing them, with documentation pointing to the preferred replacement.

• UserInteraction was superseded by InteractionCounter for tracking engagement metrics
• Map was superseded by more specific types like ParkingMap and TransitMap
• ReservationPackage was superseded by modeling multiple Reservation instances directly

Implementers should audit their structured data for superseded types and migrate to current specifications to ensure AI parsers interpret their markup using the latest semantic models and property definitions.

The process of identifying and merging disparate records that refer to the same real-world entity. When defining a Schema.org Type, entity resolution ensures a single @id consolidates all attributes, preventing the AI from interpreting one organization as two separate nodes.

• Uses probabilistic matching on attributes like name and address
• Critical for Knowledge Graph Population
• Prevents citation fragmentation in generative outputs

The determination of logical correspondences between concepts in different ontologies. This ensures that a local Product type correctly maps to the canonical Schema.org Product class, enabling interoperability across disparate systems.

• Uses sameAs and subClassOf relationships
• Bridges internal taxonomies with external vocabularies
• Essential for Vocabulary Mapping at scale

The selection of a preferred URL and structured data identifier when multiple variants exist. For a Schema.org Type, a canonical @id consolidates ranking signals and prevents the AI from treating duplicate pages as separate, competing entities.

• Uses rel=canonical in tandem with @id
• Prevents entity duplication in knowledge graphs
• Strengthens the authority of the primary type definition

The process of distinguishing between entities that share the same name by analyzing contextual clues. When defining a Schema.org Type, disambiguation ensures the AI correctly identifies the specific entity—differentiating a company from a product with the same label.

• Leverages additionalType and description properties
• Uses external identifiers like Wikidata Q-IDs
• Critical for accurate Entity Salience Optimization