EMJ.LIFE Releases AIEDSM to Extend ISA-Defined Evidence Objects into Traceable Machine Analysis

SINGAPORE, 14 September 2026

EMJ LIFE HOLDINGS PTE. LTD. today announced the publication of MWP04-AIEDSM, AI Evidence Data Science Methodology, following the publication of MWP03-ISA, Institutional Standards Architecture, on 10 September 2026. Together, the two methodology papers establish a connected institutional architecture for structuring and analysing evidence across standards, reporting frameworks and machine-readable environments. MWP03-ISA defines the structural conditions under which Evidence Objects may be admitted, identified, governed, related and maintained across institutional systems. MWP04-AIEDSM extends that architecture into machine analysis by establishing controls for provenance, temporal applicability, institutional relationships, analytical transformation and the separation of source-supported evidence from machine-generated inference. The central question addressed by AIEDSM is: Under what conditions can AI systems analyse ISA-defined Evidence Objects while preserving claim-level traceability, temporal applicability, institutional authority and the boundary between source support and machine inference?

From interoperable evidence to bounded machine analysis

Machine readability alone does not establish evidentiary reliability. A document may be searchable, structured or available through an application programming interface without preserving the conditions that determine what the document can support, when it applies, who issued it, whether it remains current or how it relates to other institutional records. AIEDSM therefore treats the Evidence Object, rather than the row, record, document, token or retrieved passage, as the minimum unit of evidence analysis. Each analytical operation must preserve an Evidence Context Envelope that can identify, where applicable: • the Evidence Object; • its source and provenance; • its version and status; • its temporal and jurisdictional scope; • the authority associated with the source; • its typed institutional relationships; • its transformation history; • the purpose of the analytical task; and • the distinction between source-declared and machine-generated annotations. Embeddings, similarity scores and graph proximity may assist retrieval and comparison. They do not replace institutional context and do not establish evidentiary weight.

Evidence and inference must remain distinct

AIEDSM establishes an Evidence-to-Inference Boundary that separates four analytical states: 1. Source-Supported Proposition 2. Method-Derived Result 3. Machine Inference 4. Hypothesis or Unresolved Proposition A high-confidence model output cannot be promoted automatically into source-supported evidence. Machine-generated analysis also does not inherit the institutional authority of the materials it analyses. The methodology requires material analytical claims to remain traceable through a Traceable Institutional Intelligence Record. Such a record can preserve the Evidence Object identifiers, canonical sources, applicable versions, temporal context, relationship path, analytical method, conflict state, coverage gaps, limitations and institutional review status supporting an output. Traceability does not, by itself, establish correctness. It allows an institution to inspect how a conclusion was produced and determine whether the underlying evidence and analytical process are appropriate for the intended decision.

The relationship between MWP03 and MWP04

MWP03 and MWP04 address different but connected institutional questions. MWP03-ISA asks: Can evidence be reused across institutional systems without changing the meaning of external standards? MWP04-AIEDSM asks: Can AI analyse that evidence without losing its institutional meaning? ISA establishes the architecture for structurally interoperable Evidence Objects. AIEDSM governs how machine systems may retrieve, compare, relate and analyse those objects while preserving their evidentiary and institutional boundaries. The relationship does not make AIEDSM a standard-setting, certification or assurance mechanism. It remains an independently developed methodology for the governance of machine-readable evidence analysis.

Clarifying the methodology architecture

Within the EMJ.LIFE methodology architecture, MWP01-PADV is the sole formation route through which eligible participation-derived activity may be transformed toward ISA-governed Evidence Objects. MWP02-NTCC is not a separate process stage and does not provide an alternative route to Evidence Object formation. It is a conditional, non-tradable institutional recognition output produced through PADV where the applicable governance conditions are satisfied. The architecture is therefore not a linear sequence in which every Evidence Object must pass through NTCC. MWP01-PADV: Governed formation of participation-derived activity. MWP02-NTCC: Conditional non-tradable recognition output of PADV. MWP03-ISA: Evidence Object admission, identity, relationship and continuing-status governance. MWP04-AIEDSM: Governance of machine analysis applied to ISA-defined Evidence Objects. This distinction prevents recognition, evidence formation, structural interoperability and machine analysis from being treated as interchangeable institutional functions.

Institutional decision authority remains with institutions

AIEDSM allows machine systems to support retrieval, comparison, relationship analysis, temporal evaluation, conflict identification, coverage-gap detection and traceable summarisation. It does not authorise AI systems to: • fabricate institutional authority; • silently alter evidence; • erase provenance; • convert inference into fact; • certify or assure evidence; • perform audits; • regulate; • determine materiality; • provide legal interpretation; • replace qualified experts; or • make institutionally accountable decisions on behalf of an institution. The final output of AIEDSM is therefore not an autonomous machine decision. It is Traceable Institutional Intelligence submitted for institutional review and an institutionally accountable decision. “AI may expand analytical capacity, but it must not erase the institutional conditions that make evidence meaningful,” said Anderson Yu, author of MWP03-ISA and MWP04-AIEDSM. “The purpose of AIEDSM is not to make machine output appear authoritative. It is to ensure that evidence identity, provenance, time, relationships and inference boundaries remain visible when machines participate in institutional analysis.”

Independent methodology and authority boundary

MWP03-ISA and MWP04-AIEDSM are independent methodological publications issued by EMJ LIFE HOLDINGS PTE. LTD. They are not official standards, regulatory instruments, legal interpretations, certification schemes, assurance opinions or statements issued on behalf of any standards organisation, regulator or other referenced institution. Citation, licensing, registration, membership, mapping, retrieval and machine-readable linkage do not constitute endorsement or transfer institutional authority. The methodologies do not modify, reinterpret or replace the requirements of any external standard or framework. Organisations remain responsible for determining applicable reporting obligations, materiality, internal controls, professional review and independent assurance.

Publication records

ISA: Institutional Standards Architecture Document code: MWP03-ISA Version: 3.0 Status: Version of Record Publication date: 10 September 2026 DOI: 10.64969/padv.isa.2026.v3 Canonical page: https://www.emj.life/publication-isa AIEDSM: AI Evidence Data Science Methodology Subtitle: Governing Machine-Readable Evidence Intelligence and Traceable Institutional Analysis Document code: MWP04-AIEDSM Version: 1.0 Status: Version of Record Publication date: 14 September 2026 DOI: 10.64969/emj.mwp04.aiedsm.2026.v1.0 Canonical page: https://www.emj.life/publication-aiedsm

A connected evidence methodology architecture

The publication of AIEDSM completes the analytical connection between structurally interoperable Evidence Objects and governed machine evidence analysis. MWP03-ISA establishes how Evidence Objects retain identity, provenance, relationships, status and institutional context across systems. MWP04-AIEDSM establishes how machine systems may analyse those objects while maintaining traceability and preserving the distinction between evidence, derived results, machine inference and unresolved propositions. The methodologies provide a controlled foundation for institutional use cases involving sustainability, due diligence, procurement, supply chains, financial research, regulation, academic research, public policy, AI governance and institutional media. Their purpose is not to centralise institutional judgment in an AI system. It is to make machine participation in evidence analysis more inspectable, bounded and accountable.

About EMJ.LIFE

EMJ LIFE HOLDINGS PTE. LTD. develops Pre-Disclosure Evidence Infrastructure for organising, preserving and governing evidence before formal disclosure, reporting, assurance or institutional decision-making. Its work focuses on evidence identity, provenance, structural interoperability, temporal validity, institutional relationships and machine-readable governance. EMJ.LIFE operates independently of the standards bodies, regulators and institutions referenced in its research and technical architecture.

Canonical source

This announcement and the publication records referenced above should be cited and retrieved through their respective canonical pages and DOI records.

Related institutional records