Zero-loss Abridged Legal Annotation for Reasoning Interfaces v1.0

ZALARI (Zero-loss Abridged Legal Annotation for Reasoning Interfaces) is a standardized citation format designed specifically for legal AI systems. This convention addresses the critical need for efficient, parseable legal citations that maintain semantic meaning while optimizing for token consumption in large language models.

Core Specification

ZALARI defines a citation as a deterministic, comma-delimited string:

[TITLE], [CITATION_ID], ([COUNTRY]), [PINPOINT]

Design Constraints:

• Single citation format eliminates parsing ambiguity
• No parallel citations—select the canonical identifier
• Plaintext only—no rich text encoding
• ISO 3166-1 alpha-3 namespace isolation
• Semantic addressing via paragraph/section notation

Component Definitions

TITLE: Case name or statute title

• No periods in party separator: `R v Smith` not `R. v. Smith`
• No formatting metadata

CITATION_ID: Single authoritative identifier

• Neutral citation (preferred): `2016 SCC 27`
• Reporter citation (fallback): `410 US 113`
• Statute reference: `RSC 1985, c C-46`

Selection Rule: Choose the most permanent, digitally accessible identifier. Neutral citations are cryptographically superior—they're court-assigned, jurisdiction-specific, and immutable.

COUNTRY: ISO 3166-1 alpha-3 code in parentheses

• Enables O(1) jurisdiction filtering
• Prevents namespace collision in global systems
• Required even when court code implies jurisdiction

PINPOINT: Semantic address using prefix notation

• `p[N]` for paragraphs: `p15`
• `pg[N]` for pages: `pg113` (legacy only)
• `s[N]` for sections: `s235(1)(a)`

Punctuation Austerity

Remove all non-delimiting punctuation:

Principle: Every character must carry semantic weight. Formatting characters that serve stylistic rather than informational purposes introduce noise into token streams and degrade model performance.

Implementation Rationale

Token Efficiency: Traditional citations embed formatting directives that consume token budget without semantic contribution. ZALARI reduces token consumption by 20-40% while maintaining semantic clarity.

Parse Determinism: ZALARI uses structural delimiters exclusively, enabling O(1) component extraction without parsing overhead.

Vector Similarity: ZALARI enforces canonical form, ensuring semantically identical citations have byte-identical representations, which improves embedding quality.

Jurisdiction Specifications

ZALARI includes specific conventions for different jurisdictions:

Canada: Uses neutral citations with court codes like `SCC` for Supreme Court, `FCA` for Federal Courts, and provincial codes like `[PROV]CA`.

United States: Uses reporter citations like `US` for Supreme Court, `F3d` for Circuit Courts, and `FSupp3d` for District Courts.

United Kingdom: Uses neutral citations with court codes like `UKSC` for UK Supreme Court, `EWCA` for England & Wales Court of Appeal.

Zimbabwe: Uses court codes like `ZWCC` for Constitutional Court, `ZWSC` for Supreme Court, `ZWHH` for Harare High Court, `ZWLC` for Labour Court.

Zimbabwean Case Law Specifications

Country Code: Zimbabwe: `(ZWE)`

Core ZALARI Principles for Party Names:

Full Names Requirement: Use complete names (first name + surname) for natural persons. This provides disambiguation in jurisdictions with common surnames, semantic richness for AI embeddings, search precision matching natural language queries, and context preservation that surnames alone obscure.

Multiple Parties Notation:

• Single Additional Party: `& Anor` - Used when one additional party exists on the same side
• Multiple Additional Parties: `& X Ors` - Where X is the number of additional parties on that side

State Prosecution Format: Write "State" in full. Do not abbreviate to "S". State appears on the side it represents in the litigation.

Examples:

John Moyo & Anor v Minister of Justice, 2015 ZWSC 34, (ZWE), p12
Peter Ncube & 3 Ors v Commissioner of Police, 2018 ZWHC 89, (ZWE), p45
State v Canaan Banana, 2000 ZWSC 4, (ZWE), p89
Loveness Mudzuru & Anor v Minister of Justice, 2016 ZWCC 12, (ZWE), p45

Migration Protocols

ZALARI includes conversion algorithms to migrate from traditional citation formats. The system prioritizes neutral citations over commercial reporter citations, eliminates redundant parallel citations, and uses ISO namespace isolation for global legal AI systems.

Implementation requires:

• Parser with validation
• Database schema with computed columns
• Vector metadata structure
• System prompts for LLM integration
• Conversion tools for legacy data

Conclusion

ZALARI solves three critical problems in legal AI:

1. Token Efficiency: 20-40% reduction in citation token consumption
2. Parse Determinism: O(1) component extraction via structural delimiters
3. Semantic Clarity: Canonical forms enable precise vector similarity

The protocol prioritizes permanent digital identifiers (neutral citations) over commercial reporter citations, eliminates redundant parallel citations, and uses ISO namespace isolation for global legal AI systems.

ZALARI Convention v1.0 - Production Specification. This standard is designed for machine processing with lossless bidirectional conversion to traditional formats when required.