ONTOLOGY PROTOCOLS
06. Concept Protocol
Revision 1.2
1. Purpose
The Concept Protocol establishes the place of a concept in the semantic architecture of the Creative Causality Research Center and the rules for its canonical representation in the evolving ontology. The document continues the architecture of Protocols 01–05 and answers the question: what exactly must be preserved when the system asserts that it is referring to the same concept regardless of language, wording of the definition, classification, and changes in knowledge about its referent. The Protocol does not determine in advance the nature of objects of Reality. It establishes the architecture for representing concepts about them.
2. Concept as a Semantic Unit
A concept is one type of semantic unit. A concept is a semantic unit by means of which the system stably distinguishes and represents definite content as repeatedly recognizable content independently of a particular linguistic designation. Not every semantic unit is automatically a concept. An assertion, a specific event, an individual observation, a model, or a complex semantic construction may also be distinguishable and addressable, but their belonging to the conceptual level must be established separately. Semantic unit is a broader category than concept.
3. Criterion of Concepthood
The basis for distinguishing a concept is the need to stably preserve an independent semantic distinction and repeatedly recognize the corresponding content in different contexts of representation. A new conceptual distinction is necessary when some content must be distinguished from other content and this distinction cannot be preserved by existing concepts without substantial semantic loss. The presence of a separate word is not sufficient grounds for the existence of a separate concept. Likewise, the absence of a separate word does not prove the absence of an independent concept. Lexical distinction ≠ semantic distinction. Semantic distinction ≠ mandatory lexical distinction.
4. Concept and Term
A concept is not identical to a term. A term is a linguistic designation by means of which a concept is expressed in a particular language or terminological system. One concept may have several terms. One term may be used to designate several different concepts. Concept ≠ Term ≠ Word. The identity of a concept must not depend on preservation of a particular verbal designation.
5. Concept and Definition
A concept is not identical to its definition. A definition is a formulated fixation of the content of a concept in a particular state of the ontology. A concept may preserve semantic identity while the wording of its definition is refined. Therefore, a change in the text of a definition must not automatically be treated as the emergence of a new concept. At the same time, preservation of the previous Concept ID is not sufficient grounds for considering a concept unchanged if its content has been substantially transformed. Concept ≠ Definition. Change of definition ≠ automatically change of concept. Preservation of identifier ≠ automatically preservation of concept.
6. Canonical Definition
For a concept that has reached the necessary degree of semantic definiteness, a current canonical definition is established. The canonical definition is the normative fixation of the meaning of the concept in the current revision of the ontology. Canonicity does not mean absolute immutability or final metaphysical truth of the definition. It means that, when interpreting the Center’s texts and data, this definition is used as primary until its justified revision. If a canonical definition has not yet been formed, this state must be explicitly indicated. A missing definition must not automatically be completed from external sources.
7. Boundaries of Meaning
A positive definition is insufficient to ensure semantic unambiguity. For a key concept, insofar as necessary, it must be specified not only what it means but also what the concept does not designate within the system. A boundary of meaning prevents conflation of the concept with nearby, historically related, everyday, or external terminological constructions. Definition establishes content. Boundary of meaning protects content from substitution.
8. Architectural Role of a Concept
For a key concept, its role in the semantic architecture must be determinable. The architectural role does not answer what function the presumed referent performs in Reality, but rather: for preservation of what semantic distinction is this concept necessary to the system? Therefore, the architectural role of a concept and the function of its referent are different characteristics. This distinction is necessary so that the structure of the representation system is not transferred onto the structure of Reality itself.
9. Concept and Referent
A concept is not its referent. A concept may refer to one presumed referent, a set of referents, a class of referents, an indeterminate referent, or content for which a real referent has not been established or is not presumed. The presence of a concept does not prove the existence of its referent. The absence of an established referent does not destroy the concept as a semantic unit. Concept ≠ Referent. Referential information must be stored separately from the canonical definition of the concept.
10. Concept and Extension
For concepts to which distinction by extension or set of referents applies, extension must not automatically be identified with the concept itself. For individual objects, relations, processes, absence, and other types of content, the concept of extension may be inapplicable or may require separate definition. Inapplicability must be explicitly indicated rather than covertly replaced with an empty value. Two concepts may have coinciding known referents while preserving different semantic content. A change in the known set of referents also does not necessarily mean a change in the concept itself. Coincidence of referents or extensions is not sufficient grounds for recognizing concepts as identical. Difference of known referents by itself is likewise not sufficient grounds for recognizing concepts as different. Concept identity is established at the level of its semantic content and essential semantic relations.
11. Concept and Assertion
A concept is not an assertion. A concept by itself does not assert that its corresponding content exists in Reality, possesses a particular property, or stands in a particular relation. Truth evaluation T/F is inapplicable to a concept as such. Truth status applies to an assertion about a state of affairs, while epistemic status applies to the degree to which the corresponding assertion is established in the knowledge system. This does not mean that a concept is automatically assigned the logical status N/A: rather, this type of evaluation belongs to a different category of construction. If a type of evaluation is inapplicable to a concept, this must be represented as inapplicability of that type of evaluation, not as epistemic status U and not as an unknown value. Concept → participates in → Assertion → receives corresponding statuses. Epistemic status must not be stored as a direct property of a concept if it actually characterizes an assertion about the concept, its referent, or a relation.
CCO 1.2 clarification. For a concept as such, truth evaluation is inapplicable: N/A is recorded in the applicability-of-truth-evaluation coordinate. This is neither a truth status nor an epistemic status of the concept. Truth status T/F, epistemic status K_T/K_F/U, and applicability of truth evaluation (applicable/N/A) are separate coordinates and must not be combined.
12. Relations of a Concept
A concept does not exist in the system in isolation. It may enter into typed relations with other concepts, assertions, models, and other addressable semantic units. Typing of such relations is performed in accordance with Protocol 05. The presence of a relation between concepts does not automatically imply the presence of a corresponding relation between their referents. Nor is every relation of a concept to other elements of the system part of its definition. Relations of a concept in the semantic architecture ≠ relations of presumed referents in Reality.
13. Essential and Contextual Relations
Relations of a concept may have different significance for preserving its semantic definiteness. An essential semantic relation is a relation without which the canonical content of the concept becomes insufficient to distinguish it from other content. A contextual relation is a relation that allows the concept to be considered within a particular domain, model, study, or projection but does not by itself determine its distinguishability as this content. This distinction must not be established arbitrarily. In the absence of sufficient grounds, the status of the relation must remain unestablished. The essentiality of a relation is an assertion about the structure of the concept and therefore requires its own grounds.
14. Concept and Classification
The place of a concept in a classification is not identical to the concept itself. A concept may be included in different admissible classification projections depending on the basis of classification. A change in classification position does not necessarily mean a change in the meaning of the concept. Place in a tree ≠ identity of concept. Classification is a way of organizing representation and must not covertly become a criterion of the ontological nature of the referent.
15. Concept and Context
Some concepts preserve their content across a broad range of contexts. Others depend substantially on domain of use, level of description, model, or historical state of terminology. The same linguistic designation in different contexts is not sufficient grounds for assigning one Concept ID. If a change of context changes essential semantic content, it must be investigated whether there is one context-dependent concept, different meanings of one term, or several independent concepts. Until this distinction is established, the system must not automatically merge the corresponding contents.
16. Semantic Identity
The identity of a concept is determined neither by its name, storage location, number, nor a particular revision of its definition. Protocol 06 establishes the subject of identity testing — semantic continuity of distinguishable content and its essential relations — but does not specify a completed algorithm for establishing that identity. Concept ID is intended to address presumed identity but does not create it. In the general machine architecture, Semantic ID is the generic designation for the class of persistent semantic identifiers; Semantic Unit ID is the identifier of an independently addressable semantic unit; Concept ID is a specialized Semantic Unit ID for a semantic unit of type Concept. These designations do not introduce three parallel identifiers for one entity. An ID records presumed identity but is not proof of its preservation. Final criteria of persistent identity, versioning, splitting, merging, and transition from change to metamorphosis must be established by subsequent protocols of machine representation and ontology development.
17. Change of a Concept
At least three types of change must be distinguished. Refinement changes the mode of expression or increases precision of the definition without an established change in semantic identity. Expansion or narrowing changes the boundaries of application of a concept and requires separate testing of preservation of its identity. Redefinition changes essential content to such an extent that recognition of a new concept may be required. This classification is not an automatic algorithm. It establishes the need to test semantic continuity. Neither an editor nor a machine system should conceal a substantial transformation of a concept as an ordinary textual change.
18. Semantic Gap
Not all characteristics of a concept must be known at the moment it is introduced. A boundary of meaning, part of its relations, referential status, historical origin, or another characteristic may remain unknown. Such absence must be designated as absence of established content. It must not automatically be filled by the most probable value, an external definition, statistical association, or familiar use of the term. The indeterminate must remain indeterminate until grounds for its determination appear. At the same time, the distinction established in Protocol 05 between substantive unknown and technical absence of an assigned characteristic is preserved.
19. Multilingual Representation
Linguistic forms are representations of a concept, not independent concepts, if sufficient semantic equivalence is preserved. Translation must therefore be performed not through mechanical replacement of words, but through correlation of the linguistic expression with the canonical content of the concept. Concept → linguistic representation₁ / linguistic representation₂ / linguistic representation₃. Translation between linguistic expressions must be controlled through the concept, not only through direct word correspondence. If another language lacks an expression that preserves essential meaning without loss, that loss must be indicated rather than hidden by an approximate translation.
20. Canonical Term
For practical work, a concept may have a canonical term in the primary language of the current ontology revision. The canonical term serves as the principal linguistic designation. It does not thereby become the concept itself. Replacement of the canonical term while preserving semantic identity does not require creation of a new concept. Historical, alternative, and translated designations may be preserved as related linguistic representations.
21. Minimum Concept Passport
The concept passport must distinguish the semantic profile of the concept from metadata of its canonical record. This is necessary so that properties of the record are not taken for properties of the concept itself. The semantic profile may include: Canonical Term; Canonical Definition; Boundary; Architectural Role; Concept Relations; Referential Information; Language Representations. Metadata of the canonical record may include: Concept ID; Sources; Version; history of changes; and other service information. This separation does not mean that Canonical Term or Language Representations are the concept itself: they remain elements of its canonical representation. The passport is a minimum working structure and does not prove that the listed fields exhaust the meaning of the concept. Epistemic status is not included as a direct property of a concept. It applies to corresponding assertions.
22. Conditions for Introducing a New Unit and a New Concept
Creation of a new addressable semantic unit must be distinguished from introduction of a new conceptual type or independent conceptual distinction. A new addressable semantic unit may be required to represent a new individual object, event, observation, or other content even when it is fully described by existing concepts. A new independent concept must not be introduced merely because a new word, term, translation, or wording variant has appeared. It requires an independent semantic distinction that cannot be preserved by existing concepts without substantial semantic loss. Before creating a new Concept ID, it is necessary to check whether a concept with this content already exists; whether the new expression is a synonym or translation; whether the distinction is only contextual; whether the presumed new concept is a composite construction of existing concepts; whether the content is sufficiently established for independent addressability; and whether a new entity is being used to fill a place that should remain indeterminate. The number of concepts is not a measure of ontology quality.
23. Splitting Concepts
If one linguistic designation is used for several substantially different contents, these contents must not remain under one Concept ID merely for lexical unity. In such a case, concept splitting is required. Splitting must be accompanied by a record of the grounds for distinction and preservation of the history of previous use. Homonymy, polysemy, and historical change of a term must not become hidden conflation of semantic units.
24. Merging Concepts
If two previously distinguished concepts later prove to be semantically identical, there may be grounds for merging them. However, coincidence of names, individual definitions, classification positions, or known referents is insufficient for automatic merging. Before merging, content, boundaries of meaning, essential semantic relations, referential characteristics, history of the concepts, and the reason for their original distinction must be checked. If identity is not established, the distinction is preserved.
25. Source and Authorship of a Definition
The source of a concept, the source of its definition, and the source of assertions about its referent may be different. They must therefore not be stored as one undifferentiated field. Authorship of wording does not establish the truth of assertions about Reality. However, the origin of a definition is important for reconstructing the history of semantic development and checking how the current canonical content arose. History of origin is part of ontology transparency but does not replace semantic analysis.
26. Machine Interpretation of a Concept
A machine system must not determine the meaning of a concept solely by its name, frequency of co-occurrence of words, or nearest statistical association. Where a canonical record exists, explicitly established elements of the concept’s semantic architecture take priority. A machine must not complete absent elements of the passport as though they already belonged to canonical content. A probabilistic hypothesis about meaning may exist only as a separately marked hypothesis and must not covertly alter the concept itself. Interpretation ≠ canonization.
27. Architectural Openness
The proposed structure of a concept is not declared final. If further research discovers a characteristic without which an essential semantic distinction cannot be preserved, the architecture of concept representation may be expanded. However, a new field or relation type must not be introduced merely because it is technically convenient or common in existing ontological systems. Expansion must have semantic grounds. The architecture of concept representation exists for preservation of meaning, not meaning for preservation of the architecture.
28. Fundamental Limitations
The presence of a word does not imply the presence of a separate concept. The presence of a concept does not imply the existence of its referent. Coincidence of terms does not imply identity of concepts. Difference of terms does not imply difference of concepts. Coincidence of known referents does not imply identity of concepts. Change of known referents does not automatically imply change of concept. Change of definition does not automatically imply emergence of a new concept. Preservation of Concept ID does not automatically imply preservation of semantic identity. Classification position does not imply the nature of the referent. Absence of a definition does not imply a right to complete it. Machine interpretation does not imply canonical meaning.
29. Canonical Principle
A concept is not a word, not a definition, not a referent, and not an assertion. A concept is a semantic unit by means of which the system stably distinguishes and represents definite content as repeatedly recognizable content independently of a particular linguistic designation. Terms represent the concept in language. A definition fixes its content in a particular state of the ontology. Boundaries of meaning protect the content from semantic substitution. The architectural role shows for preservation of what distinction the concept is necessary to the system. Relations connect it with other semantic units. Reference correlates it with a presumed referent. Assertions make it possible to speak about the concept, its referent, and related states of affairs. An identifier provides addressability of presumed semantic identity but does not create that identity. Therefore, a change of term, definition, classification, known extension, or other representations of a concept must not automatically be taken either as preservation or as loss of the concept itself. Protocol 06 determines what exactly must be tested when establishing semantic continuity, but the final procedure for identity, versioning, splitting, merging, and metamorphosis is established by subsequent protocols. When grounds are insufficient, the question of identity must remain open.
← 05. Ontological Graph Deployment Protocol · Ontology Protocols