Conceptual dependency theory¶
Normalize paraphrastic natural-language inputs into canonical, language-independent event structures built from primitive acts, typed participants, dependencies, and inference-enabling causal expectations.
Core Idea¶
Conceptual dependency theory is Roger Schank's historical knowledge-representation framework for expressing the meaning of natural-language events in a canonical form intended to be independent of the particular words used. It decomposes actions into a small inventory of conceptual primitives, links them to actors, objects, directions, instruments, times, locations, and states, and uses dependency relations to support inference. Paraphrases should converge on one underlying representation when their asserted meaning is the same.
A parser identifies an event and selects a primitive act such as ATRANS for transfer of an abstract relation, PTRANS for physical movement, MTRANS for information transfer, PROPEL for applied force, or INGEST for taking something into an animate object.
Scope of Application¶
The abstraction is literal wherever practitioners can identify the same constitutive roles, apply the same boundary tests, and obtain the same kind of output. The following habitats are uses of Conceptual dependency theory itself, not metaphors based only on resemblance.
- Natural-language parsing. Mapping multiple verb forms and sentence structures into canonical event representations.
- Paraphrase handling. Treating semantically equivalent inputs alike despite lexical differences.
- Inference. Deriving possession, location, state-change, and causal expectations from represented acts.
- Question answering. Answering from normalized event roles rather than matching surface words alone.
- Narrative understanding. Connecting event representations with memory structures and later script theory.
- History of AI. Studying a knowledge-intensive alternative to syntax-only and search-oriented processing.
Clarity¶
A clear account of Conceptual dependency theory must preserve the recognition invariant stated in the Core Idea rather than rely on the title alone. Identify the publication or implementation because primitive inventories and notation changed over time. Separate the input sentence, asserted meaning, chosen primitive, filled roles, and licensed inferences. Do not label a word-dependency parse or arbitrary semantic graph as conceptual dependency. Evaluate representational coverage, inference correctness, and parsing performance as separate questions.
Manages Complexity¶
Conceptual dependency theory manages complexity by replacing a diffuse field of observations or possible operations with a bounded role structure: natural-language input supplies a sentence or discourse supplies a surface whose asserted event meaning is to be represented.; canonicalization goal supplies paraphrases with the same meaning are directed toward one underlying form.; primitive acts supplies a small typed vocabulary such as ATRANS, PTRANS, MTRANS, and PROPEL decomposes events.; participant roles supplies actors, objects, recipients, sources, destinations, and instruments fill declared slots.; state and setting supplies time, location, possession, physical state, and mental state constrain the event..
Abstract Reasoning¶
- Determine the asserted event meaning and resolve only the ambiguity justified by context. 2. Select the conceptual primitive or composition of primitives matching the event transition. 3. Bind actors, objects, directions, recipients, instruments, times, locations, and relevant states. 4. Normalize paraphrases by checking whether the resulting role and transition structure is identical. 5. Apply only the inference rules attached to the selected primitives and dependencies.
Knowledge Transfer¶
The strict upward abstraction is Representation. Conceptual Dependency Theory instantiates Representation because it encodes event meaning in a structured surrogate whose primitives and relations make some inferences directly available while suppressing surface wording. Within canonical event semantics, the full mechanism transfers literally when the same roles and boundary tests recur. Beyond that domain, only the parent-level skeleton should travel. Reusing the label Conceptual dependency theory after removing its constitutive vocabulary would hide a change of mechanism behind an analogy. The honest transfer rule is therefore two-stage: recognize the domain-specific pattern first, then lift only the parent relation that remains invariant under a substrate change.
Relationships to Other Abstractions¶
Current abstraction Conceptual dependency theory Domain-specific
Parents (1) — more general patterns this builds on
-
Conceptual dependency theory is a kind of Representation Prime
Conceptual Dependency Theory instantiates Representation because it encodes event meaning in a structured surrogate whose primitives and relations make some inferences directly available while suppressing surface wording.
Hierarchy path (1) — routes to 1 parentless root
- Conceptual dependency theory → Representation → Abstraction
Neighborhood in Abstraction Space¶
Conceptual dependency theory sits in a sparse region of the domain-specific corpus (91st percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.
Family — Unclustered & Miscellaneous (1565 abstractions)
Nearest neighbors
- Language Production — 0.80
- Semantic parsing — 0.78
- Functional discourse grammar — 0.78
- Logic form — 0.78
- Functional Generative Description — 0.78
Computed from structural-signature embeddings · 2026-09-08