Knowledge mapping
Durable research-point identifiers, concept and alias resolution, evolving research umbrellas and prospective relationships.
Multilevel, Interdisciplinary, Longitudinal Evaluation of Systems — a proposed research architecture for discovering, relating and critically evaluating research points across disciplinary, institutional and historical boundaries.
A research point should remain discoverable even before all relevant disciplines or relationships are known. MILES proposes to record individual claims, questions, findings and events; connect them to multiple subject umbrellas; and independently test those connections without treating thematic proximity as causal proof.
Overarching question: How can interdisciplinary knowledge, historical evidence and institutional relationships be systematically discovered, connected, evaluated and revised across time to improve the completeness, transparency and reliability of research and policy reasoning?
Durable research-point identifiers, concept and alias resolution, evolving research umbrellas and prospective relationships.
Source and relationship-level provenance, alternative explanations, contradictions and explicit evidence status.
Historical sequences, institutional power, delayed effects, feedback, adaptation, possible causal paths and outcomes.
These layers can operate iteratively: discovering a new contradiction can reopen a mapped relationship or change an analytical conclusion.
| Field | Purpose |
|---|---|
| Research-point ID | Persistent identifier with precise claim, question, observation or event. |
| Concept resolution | Canonical labels, aliases, distinct but related concepts and discipline-specific definitions. |
| Research umbrellas | Multiple classifications which can expand without excluding newly identified points. |
| Temporal and institutional context | Effective period, date of event, publication date, geographic and organisational boundaries. |
| Relationship-level provenance | Each edge has its own identifier, direction, type, sources and epistemic status. |
| Review history | Supporting and contradicting evidence, limitations, corrections and attributable revisions. |
For example, underemployment, skills underutilisation and precarious work may connect without being considered synonyms. A proposed edge is not proof of causation.
A cross-layer module records policy and research inflection points: announcement, enactment, commencement, implementation, observed outcomes, unintended consequences and possible system “snap-back.” It also records methodological discoveries, review decisions and corrections.
Cabaj's systems-change distinction between strategic learning, systems change and mission impact informs evaluation of results. These types of results are distinct from macro, meso and micro levels of analysis.
The proposed study considers employment, education, welfare, migration, technology, equality, finance and economic mobility, examining possible interactions rather than presuming a single cause. The 2007 baseline precedes the financial crisis; 2027 is a prospective follow-up period as of November 2026.
Employment structures, financial conditions and institutional arrangements.
Financial disruption, workforce restructuring and policy responses.
Outsourcing, gig markets, zero-hours arrangements, pay and occupational progression.
Furlough, support eligibility, employer decisions and disruptions.
Inflation, skills recognition, recruitment, migration policy and changing institutions.
New evidence, policy outcomes, revisions and comparison with earlier expectations.
Explore disciplines, topic umbrellas, methods and potential connections.
Questions, coverage, provenance, research gaps and contradiction assessment.
Knowledge networks, provenance and discovery across distributed public information.
Related proposed AI capabilities include historical as-of-date retrieval, claim-to-evidence coverage, contradiction detection, cross-conversation evidence reuse and contributor attribution. These remain proposals or independently documented developments, not capabilities asserted to be operational here.
UK Labour Market Analysis → · Research Collaborations → · Research Overview →
Review against realist evaluation, process tracing, system dynamics, longitudinal econometrics and existing knowledge-graph approaches including OpenAlex, Wikidata and nanopublications. Their prior contributions must be credited; MILES' additional usefulness remains to be demonstrated through comparative tests.
External reviewers are invited to examine disciplinary completeness, concept resolution, source independence, causal validity, reproducibility, missing mechanisms and the feasibility of a bounded pilot.
Proposed pilot: Trace one documented UK employment-policy sequence using persistent point IDs, independently evidenced edges and dated milestones. Distinguish observed changes from contested explanations and hypotheses.
This working record shows how published research informs MILES, which parts are already established elsewhere, and what remains untested. It separates support for underlying methods from evidence that the combined MILES architecture is distinctive or effective.
| Source and date | What it contributes | MILES component | Remaining question |
|---|---|---|---|
| Swanson (1987), Two medical literatures | Literature-based discovery across apparently separate fields. | Knowledge mapping; unexpected connections. | Does an extensible point map discover useful links that standard searches miss? |
| Open Research Knowledge Graph and associated research | Structured scholarly contributions and research comparisons. | Persistent research points; structured comparison. | What integration is not already offered by research knowledge-graph systems? |
| Chhetri et al. (2025), evidence/provenance ontology survey | Existing models for assertions, evidence and provenance. | Claim and independently evidenced relationship records. | Which existing ontologies should be reused for interoperability? |
| Systems- and Complexity-Informed Evaluation (2021) | Interrelationships, perspectives, boundaries and dynamics. | Systems analysis; institutional boundaries. | How can these constructs be operationalised rather than only diagrammed? |
| Critical integrative review of systems change (2025) | Comparative systems-change literature, including attention to power and resources. | Institutional power; distributive outcomes. | What additional evidence is required to measure cumulative disadvantage? |
| Mahoney (2012), The Logic of Process Tracing Tests | Tests of within-case historical causal explanations. | Causal inference and competing explanations. | Can researchers distinguish plausible sequences from supported causal pathways? |
| Kejriwal (2026), Semantics for Science | Evidence-centred scientific assertions, uncertainty and provenance; close conceptual prior art. | Knowledge and evidence layers. | Does MILES add a demonstrable longitudinal institutional-evaluation capability? |
| Cabaj (2019), Evaluating Systems Change Results | Strategic learning, systems-change outcomes, mission impact and adaptive responses. | Milestone evaluation; system adaptation. | Can intermediate shifts and reversion be reliably recognised? |
OpenAlex, Wikidata, nanopublications, the Open Research Knowledge Graph and evidence-centred research graphs already provide important parts of the proposed functionality. In particular, individual assertions, qualifiers, references, provenance and structured scholarly comparisons are not unique inventions of MILES. A distinct contribution would require demonstrated utility from connecting those capabilities to institutional power, dated policy sequences, contradictory evidence and reproducible systems analysis.
Scope: Begin with one bounded, documented employment-policy sequence, using 2007 as a pre-crisis baseline where supported by data. Assess evidence available by 8 November 2026; treat 2027 as a future follow-up, not observed history.
The architecture is an unpublished working concept. Neither superior discovery performance nor causal validity has been demonstrated. Potential risks include false graph links, category drift, duplicate concepts, incomplete source coverage, hindsight bias, excessive complexity and conflation of institutional association with responsibility. Independent academic assessment, technical prototyping and a reproducible pilot are still required.
Next milestone: Request interdisciplinary criticism on novelty, ontology reuse, causal validity and the minimum feasible benchmark. Publish any response or correction with its own date; do not describe an enquiry as endorsement.