Skip to main content

Glossary

Quick definitions for the terms used across the site. Cross-linked from concept pages so you can look up "what does grounded mean here?" without reading a whole page.

Argumentation primitives​

Argument​

A node in an argumentation framework. The library treats arguments as opaque IDs by default (Dung's abstract framework). Schemes and ASPIC+ give arguments internal structure (premises + conclusion).

Attack​

A directed edge in an argumentation framework — (attacker, target) — meaning the attacker's content undermines the target. May be classical (binary) or weighted (with a strength).

Support​

The dual of attack — (supporter, supported) — meaning the supporter's content reinforces the target. Lives in the argumentation-bipolar crate.

Framework​

A pair (A, R) of arguments and attacks (Dung). Bipolar adds supports; weighted adds attack weights; valued adds value promotions.

Scheme​

A named pattern of human argument (e.g., expert opinion, cause-to-effect). Schemes carry premise slots, a conclusion template, and critical questions. The argumentation-schemes crate ships ~60 schemes from Walton, Reed & Macagno (2008).

Critical question​

A standardised challenge to a scheme. E.g., for argument-from-expert-opinion: "Is the expert reliable?" "Is the field within their expertise?" Each scheme carries its own list. The bridge translates critical questions into attacks.

Acceptance semantics​

Conflict-free​

A set of arguments where no member attacks another member. The minimal property of any extension.

Admissible​

A conflict-free set that defends each of its members against external attacks (for every attacker b of a member a, some member of the set attacks b).

Grounded extension​

The unique smallest admissible set — equivalently, the least fixed point of the characteristic function. Always exists. Polynomial to compute. The "skeptical answer" — what survives if you refuse to take sides on any disputed point.

Preferred extension​

A maximal admissible set — admissible, and you can't add more without violating admissibility. Multiple preferred extensions can co-exist for the same framework. Always exists.

Complete extension​

An admissible set that contains every argument it defends. Sits between admissible and preferred. Both grounded and preferred extensions are complete.

Stable extension​

A conflict-free set that attacks every argument outside it. Stricter than preferred. Doesn't always exist.

Credulous acceptance​

An argument is in some preferred extension. The "could a reasonable observer accept this?" question. Used by the encounter bridge for proposer-side scoring.

Skeptical acceptance​

An argument is in every preferred extension. The "what survives every reading?" question. Stricter than credulous; always a subset.

Defeat​

In a value-based framework, an attack that has not been filtered out by the audience. The defeat graph is the audience-conditioned subgraph of the original attack graph.

Weighted argumentation​

Weight​

A number w ∈ [0, 1] attached to an attack edge. Higher = stronger attack.

Budget (β)​

A threshold β ∈ [0, 1] on attack weight. Attacks with w > β bind; attacks with w ≤ β drop. Single-knob "scene intensity" dial.

β-residual framework​

The Dung framework you get by removing all attacks with w ≤ β. Acceptance semantics run on the residual.

Binding / dropping​

An attack binds when w > β (acts as a Dung attack). An attack drops when w ≤ β (treated as absent).

Value-based argumentation​

Value​

Something an argument can promote (e.g., Value::new("life")). The argumentation-values crate uses string-typed values; consumers can adopt any taxonomy.

Value assignment​

A map from arguments to the set of values each promotes. Multi-value supported per Kaci & van der Torre (2008).

Audience​

A strict partial order over values, represented as ranked tiers. Each character can have their own audience; consensus is computed via MultiAudience.

Pareto-defeating​

The multi-value defeat rule: A defeats B iff for every value B promotes, some value A promotes is not strictly less-preferred under the audience. Reduces to Bench-Capon (2003) when each argument promotes one value.

Subjective acceptance​

An argument is accepted by some total ordering of the value set. NP-complete in general; capped at 6 values.

Objective acceptance​

An argument is accepted by every total ordering of the value set. co-NP-complete in general; capped at 6 values.

Encounter bridge​

Proposer​

The actor whose turn it is to propose an affordance in a scene beat. Gets scored via ActionScorer.

Responder​

The actor evaluating the proposer's affordance for acceptance. Gets queried via AcceptanceEval.

Affordance​

A scene-engine concept (from encounter): a candidate action the proposer could take. The bridge treats each affordance as backed by a scheme instance via ArgumentKnowledge.

Affordance key​

The canonical (actor, affordance_name, bindings) triple used for forward-index lookup in the state. See AffordanceKey.

Scheme instance​

A scheme with concrete bindings (e.g., expert-opinion with expert=alice, domain=military, claim=fortify_east). The result of Scheme::instantiate(&bindings).

Scheme strength​

A property of a scheme (Strong/Moderate/Weak). The SchemeActionScorer boost is proportional to strength × per-character preference_weight.

Preference weight​

A per-character, per-scheme-instance scalar in [0, 1] indicating how strongly that character holds the scheme position. Read by SchemeActionScorer.

Argument knowledge​

A trait that supplies per-character argument positions (which scheme they invoke for which action, with what bindings, with what preference weight). StaticKnowledge is the default impl for fixtures.

Error latch​

The bridge's design for handling internal failures: append errors to a Mutex<Vec<Error>> on the state, return permissive defaults, drain via state.drain_errors() after resolution.

Library / workspace​

Crate​

A Rust library/binary unit. The argumentation workspace has 7 crates: argumentation, argumentation-bipolar, argumentation-weighted, argumentation-weighted-bipolar, argumentation-schemes, argumentation-values, encounter-argumentation.

Catalogue​

A CatalogRegistry of SchemeSpecs. The default catalogue (default_catalog()) ships ~60 Walton schemes.

ENUMERATION_LIMIT​

A hard cap on framework size for subset-enumeration semantics. The core argumentation crate uses 22 (preferred / stable). The argumentation-values crate uses 6 for subjective/objective acceptance.

See also​