Part of the WebMEM® Protocol
Location: /protocol/fragments/
Status: Current Draft
Last Updated: 2026-08-24
What is a WebMEM Fragment?
A WebMEM Fragment is a modular knowledge object within a Semantic Data Template (SDT).
Each fragment represents a defined unit of publisher knowledge with a specific semantic purpose. A fragment may contain facts, definitions, explanations, indexes, policies, procedures, identity information, or other structured knowledge depending on its fragment class.
Fragments allow the machine-oriented knowledge associated with a web resource to be divided into coherent, independently identifiable objects rather than expressed as one monolithic data structure.
Together, the WebMEM fragments associated with a web resource form its Semantic Data Template.
Fragments Within the SDT
The Semantic Data Template is the page-level machine knowledge representation in WebMEM.
The relationship is:
Web Resource → Semantic Data Template → WebMEM Fragments
A single SDT may contain one fragment or many fragments. The number and classes of fragments depend on the knowledge represented by the resource.
For example, a data-rich resource might use separate fragments to represent:
- the identity and context of the resource;
- source data and factual values;
- statistics derived from underlying data;
- indexes of related or resolvable entities;
- definitions required to interpret domain-specific terminology;
- explanatory knowledge;
- and other knowledge objects appropriate to the resource.
The fragments remain distinct knowledge objects while sharing the broader context established by the SDT and the web resource in which they are published.
Why WebMEM Uses Fragments
Real-world knowledge rarely fits cleanly into a single undifferentiated object.
A web resource may simultaneously describe an entity, publish factual data about it, calculate statistics from that data, define specialized terminology, explain relationships, provide procedural information, and identify other resources connected to the subject.
WebMEM fragments provide boundaries between those different knowledge objects.
This modular approach allows publishers to:
- Separate knowledge by purpose. Different types of knowledge can be represented by the fragment class designed for that purpose.
- Preserve context. Each fragment can identify the entity, subject, dataset, scope, or other context required to interpret its contents.
- Preserve provenance. Knowledge can remain connected to the source information from which it originated.
- Represent complex resources. Multiple fragments can work together to describe different aspects of the same resource without collapsing everything into one structure.
- Support resolution. Fragments can expose identifiers, indexes, relationships, and other knowledge that helps machines move toward more precise information.
- Support validation. Fragment class and structure provide a basis for determining whether a published knowledge object conforms to the WebMEM Protocol.
Fragment Classes
Every WebMEM fragment belongs to a defined fragment class.
A fragment class identifies the semantic purpose of the knowledge object. The class communicates what kind of knowledge the fragment represents and establishes the structure and requirements appropriate to that type of knowledge.
The WebMEM Protocol currently defines 15 fragment classes.
| Fragment Class | Purpose |
|---|---|
| DataFragment | Represents structured factual data associated with an entity, subject, record, or defined context. |
| DerivedStatsFragment | Represents statistics, calculations, aggregates, or other values derived from underlying source data. |
| IndexFragment | Represents an index or collection of identifiers, entities, records, or other objects associated with a defined context. |
| DefinedTermFragment | Represents a defined term and the structured information required to establish its meaning within a domain or glossary context. |
| FAQFragment | Represents structured question-and-answer knowledge. |
| MetaFragment | Represents metadata and contextual information associated with a resource, entity, dataset, or SDT. |
| ExplainerFragment | Represents explanatory knowledge intended to clarify a concept, subject, relationship, or factual context. |
| DirectoryFragment | Represents a structured directory or collection of entities and associated information. |
| EligibilityFragment | Represents eligibility requirements, conditions, criteria, or qualification logic. |
| ProcedureFragment | Represents a defined procedure, process, sequence, or set of actionable steps. |
| GlossaryFragment | Represents a structured collection of domain terms and their definitions or semantic relationships. |
| PersonaFragment | Represents a defined persona, audience, user type, or contextual profile used to establish interpretive or application context. |
| PolicyFragment | Represents a policy, rule set, requirement, restriction, or authoritative policy statement. |
| RecommendationFragment | Represents a structured recommendation together with the context required to interpret or apply it. |
| IdentityFragment | Represents canonical identity information used to establish what an entity is and how it can be consistently identified. |
Fragment Class Is Semantic
A fragment class is not merely a formatting choice.
The class identifies the semantic role of the fragment within the SDT.
A DataFragment, for example, represents factual data. A PolicyFragment represents policy knowledge. A ProcedureFragment represents procedural knowledge. An IdentityFragment establishes identity.
Although different fragment classes may share common structural elements, their purposes are not interchangeable.
The fragment class allows the publisher to state explicitly what kind of knowledge object is being published rather than requiring a consuming system to infer that role from the contents alone.
Fragment Scope
Every fragment exists within a defined context.
Depending on the fragment class and application, that context may include:
- an entity;
- a geographic location;
- a product or service;
- a dataset;
- a record;
- a collection;
- a term or glossary;
- a policy domain;
- a procedure;
- a time period;
- or the web resource itself.
Fragment scope is important because a factual value or statement can change meaning when separated from the entity or context to which it belongs.
WebMEM therefore treats context as part of the knowledge representation rather than assuming that machines will reconstruct it from surrounding prose.
Fragment Identity
A WebMEM fragment should be independently identifiable within its SDT.
Stable fragment identity allows other WebMEM objects to reference a fragment, establish relationships with it, associate provenance with it, or use it as part of a resolution path.
Fragment identity is distinct from entity identity.
An entity identifies the real-world or conceptual object being described. A fragment identifies a particular knowledge object published about that entity or context.
Multiple fragments may therefore describe different aspects of the same entity while retaining their own fragment identities.
Fragments and Machine Fact Objects
Some WebMEM fragment classes contain explicit factual units.
WebMEM represents those atomic factual units as Machine Fact Objects.
The relationship is:
SDT → Fragment → Machine Fact Object
A fragment establishes the semantic purpose and context of the knowledge object. Machine Fact Objects represent individual factual assertions within that object where appropriate.
Not every fragment is merely a collection of facts. Some fragment classes represent higher-order knowledge structures such as procedures, explanations, policies, directories, definitions, or recommendations.
Machine Fact Objects therefore operate within the fragment model rather than replacing it.
Read the Machine Fact Objects specification →
Fragments and Provenance
WebMEM fragments may identify the source and lineage of the knowledge they contain.
Provenance can establish where information originated, which dataset or source supports it, whether values were derived, and other information necessary to understand the evidentiary basis of the published knowledge.
Where multiple facts within a fragment share the same source, provenance information may be established at the fragment level and referenced by the relevant knowledge objects.
Where facts originate from different sources, more specific provenance relationships may be expressed.
Read the Provenance specification →
Fragments and Identity
Fragments may describe or reference entities with stable identifiers.
Entity identity allows knowledge published in one fragment to remain connected to the same entity when that entity appears elsewhere in the SDT, on another web resource, or within a resolution path.
The IdentityFragment provides a dedicated fragment class for representing canonical identity information when an explicit identity object is required.
Identity and identifier behavior across WebMEM are defined separately from fragment serialization.
Read the Identity specification →
Fragments and Relationships
Fragments do not exist in isolation.
A WebMEM resource may express relationships among fragments, entities, facts, collections, source records, or other resources.
These relationships allow a machine to understand how separately represented knowledge objects participate in a larger information structure.
Relationships may also participate in machine resolution by connecting one known object to another object that can be identified or resolved more precisely.
Read the Relationships specification →
Fragments and Resolution
WebMEM fragments can participate in resolution by exposing knowledge that connects a machine’s current context to another identifiable information object.
An IndexFragment, for example, may enumerate the identifiers available within a collection or context. An IdentityFragment may establish the canonical identity of an entity. Other fragment classes may contain references or relationships that connect their knowledge to additional resources.
The WebMEM Resolution Model defines how identities, identifiers, relationships, indexes, canonical resources, and resolver structures work together to make publisher knowledge navigable by machines.
Read the WebMEM Resolution Model →
Fragment Serialization
The WebMEM fragment model is conceptually separate from its serialization.
A fragment defines what the knowledge object is. Serialization defines how that object is represented within a published resource.
The current WebMEM Protocol uses HTML-in-HTML as its primary web serialization method. WebMEM fragments are represented using structured HTML within inert <template> elements so the machine-oriented knowledge layer can travel with the human-facing resource.
This separation allows the fragment model to remain stable even if additional serialization methods are introduced in the future.
Read the HTML-in-HTML Serialization specification →
Fragment Conformance
A conforming WebMEM fragment must identify its fragment class and satisfy the requirements defined for that class.
Conformance may include requirements related to:
- fragment identity;
- required and optional properties;
- entity or contextual scope;
- field structure;
- provenance;
- relationships;
- and serialization.
Class-specific requirements are defined by the WebMEM Protocol and evaluated through WebMEM conformance and validation rules.
Read the Conformance specification →
WebMEM Fragment Classes
- DataFragment
- DerivedStatsFragment
- IndexFragment
- DefinedTermFragment
- FAQFragment
- MetaFragment
- ExplainerFragment
- DirectoryFragment
- EligibilityFragment
- ProcedureFragment
- GlossaryFragment
- PersonaFragment
- PolicyFragment
- RecommendationFragment
- IdentityFragment
Each class has a defined semantic purpose within WebMEM. The transition from WebMEM 1.x to the current protocol changes how fragments are serialized and extends the architecture toward machine resolution; it does not discard the established fragment-class model.