Rule Suspension

Suspension is fundamentally a non-destructive operation. Unlike norm abrogation, which permanently removes a rule's validity, suspension places a rule in a dormant state. The rule remains part of the normative corpus and can be reactivated automatically when the suspending condition ceases. This is critical for modeling force majeure clauses, emergency statutes, and transitional provisions where the rule's authority is merely paused, not extinguished.

Suspension is always bounded by a defined scope of application. The deactivation applies only within a specific factual context, jurisdiction, or class of agents. This implements the lex specialis principle computationally:

• A general rule remains active for all standard cases
• The suspension carves out a precise exception domain
• Outside the exception domain, the rule retains full force This scoping prevents the cascade of unintended deactivations that plague naive override systems.

Every suspension operation carries explicit temporal parameters. A rule may be suspended:

• Until a specific calendar date
• For the duration of a declared state of emergency
• While a specified condition evaluates to true This temporal logic integrates with deadline reasoning and contrary-to-duty obligation structures. When the temporal bound expires, the rule automatically reactivates without requiring a new enactment, maintaining the system's normative continuity.

Suspension interacts with the normative hierarchy graph in a structured manner. A rule from a superior authority can suspend an inferior rule, but not vice versa. The suspension operation itself inherits the authority level of the suspending rule. This ensures that lex superior derogat inferiori is respected even during exception handling, preventing lower authorities from improperly disabling mandates from higher sovereign sources.

Suspension serves as a primary mechanism for conflict preemption. When a normative collision is detected between a general rule and a specific exception, suspension resolves it by temporarily nullifying the general rule's effect within the exception's scope. This is distinct from normative repair operators that permanently alter rule text. Suspension preserves the original rule's integrity while achieving conflict-free reasoning through runtime deactivation.

Upon reactivation, the system must perform a normative entailment check to ensure the restored rule does not introduce new contradictions. The reactivation protocol typically includes:

• Re-evaluating the rule's applicability conditions
• Checking for conflicts with rules enacted during the suspension period
• Updating the normative coherence metric This verification step is essential for maintaining a conflict-free subset of active norms at all times.

The foundational legal principle that a specific rule overrides a general rule, forming the logical basis for exception handling. In computational terms, this is implemented by checking the scope intersection of two conflicting norms; if one rule's applicability conditions are a strict subset of the other's, the more specific rule is activated, and the general rule is suspended for that context. This prevents a general prohibition from blocking a specifically permitted action.

A mode of inference where conclusions are tentative and can be retracted when new evidence or superior rules are introduced. Unlike classical logic, defeasible reasoning allows a system to draw a valid conclusion from a general rule (e.g., 'contracts are binding') and then retract that conclusion when an exception is triggered (e.g., 'the signatory is a minor'). This is the core engine behind non-monotonic legal AI.

The algorithmic process of scanning a normative corpus to identify direct collisions between deontic modalities. A conflict is flagged when two active rules assign incompatible statuses to the same action:

• Obligation vs. Prohibition: One rule mandates action A, another forbids it.
• Obligation vs. Permission: One rule mandates A, another permits omitting A. Detection engines use deontic logic tensors to represent these states and compute collision matrices for resolution.

A computational strategy for resolving an inconsistent rule base by identifying the largest possible subset of rules that contains no contradictions. When a conflict is detected, the system generates multiple candidate subsets, ranks them by a normative coherence metric, and selects the optimal conflict-free set. This approach is essential for generating coherent legal advice from a body of law that contains historical contradictions or overlapping amendments.

A complex deontic construct specifying what an agent is obligated to do after violating a primary obligation. For example: 'You must not breach a contract; but if you do breach, you must pay damages.' Modeling this in AI requires temporal reasoning and normative repair operators to handle the secondary obligation without invalidating the primary prohibition. This is a key challenge for building realistic compliance systems.

A directed acyclic graph (DAG) that encodes the precedence relationships between all rules in a legal system. Nodes represent individual rules, and directed edges represent 'overrides' based on:

• Lex Superior: Constitutional law overrides statute.
• Lex Posterior: A newer statute overrides an older one.
• Lex Specialis: A specific rule overrides a general one. The graph is traversed at inference time to deterministically resolve any conflict before generating a final output.