Method-as-Metadata¶

The cataloging move of recording how a resource was produced or what procedure it employs as a first-class, controlled-vocabulary field — orthogonal to topic, creator, date, and form — so a corpus becomes filterable by technique without reading each item.

Core Idea¶

Method-as-metadata is the descriptive-cataloging move of recording how a resource was produced or what procedure it employs as a first-class, queryable metadata field on the resource record — distinct from what the resource is about (topic), who made it (creator), when it appeared (date), or what form it takes (format). Operationalized as Dublin Core's instructionalMethod for learning resources, study_design controlled-vocabulary fields in clinical trial registries such as ClinicalTrials.gov, and methodology facets in dataset catalogs on Zenodo, Kaggle, and HuggingFace. The structural move is to lift procedure to a searchable axis so that users can filter a corpus by technique — "find all studies using randomized controlled trial," "find all materials taught via inquiry-based learning," "find all datasets collected via random-digit dialing" — without reading each resource to discover its method.

The pattern applies wherever a catalog serves users whose retrieval need is technique-compatibility rather than topic-match: a systematic-review author needing all RCTs on a condition; a teacher wanting only simulation-based pedagogy materials on a topic; an auditor aggregating governance records by decision procedure. Without the field, method is buried in free-text descriptions and extractable only by reading; with it, method becomes a filter dimension alongside topic, date, and form. The field's quality depends on the specificity of the controlled vocabulary (a flat "survey" tag collapses telephone surveys, web surveys, and randomized address-frame surveys; a hierarchical vocabulary distinguishes them) and on disciplined tagging at cataloging time.

Structural Signature¶

Sig role-phrases:

the designated method field — a first-class descriptive axis recording how a resource was produced or what procedure it employs, parallel to topic, creator, date, and form
the controlled method vocabulary — the (often hierarchical) value list the field draws from (RCT, survey, inquiry-based, lecture)
the per-resource tag — the method value assigned to each record at cataloging time
the orthogonality to topic — the structural commitment holding "what it is about" apart from "how it was made," two things free-text fuses
the technique-compatibility query — the retrieval affordance the axis enables: constrain method to a value, independent of topic/date/form
the corpus-to-query collapse — the payoff: "read every item to find the technique" becomes a one-axis filter, newly supporting aggregation and crosswalking
the granularity parameter — the single design knob setting resolving power: a flat "survey" tag under-cuts (telephone/web/address-frame collapse to one), a hierarchical vocabulary resolves to the precise technique
the tagging-discipline check — consistency of application at description time, the second condition on whether the axis carries weight or merely appears to
the static-descriptor limitation — a tag about a process, not the process itself (enacting it is event-centered modeling); off cataloging the move is just tagging, the force living with index/classification parents

What It Is Not¶

Not a topic or subject descriptor. Method-as-metadata records how a resource was produced or what procedure it employs, an axis orthogonal to what it is about. A technique-filter ("find the randomized controlled trials") is a different query from a subject-filter ("find the studies on this condition"); free-text description fuses the two, and the whole point is to hold them apart.
Not a particular method value. "Inquiry-based" or "randomized controlled trial" are values the field can take; method-as-metadata is the move of having a method axis at all. Reasoning about the dimension is distinct from confusing it with one of its entries — the prime-level move is the searchable axis, not any single technique it can name.
Not the methodology itself. The field is a static descriptor about a process — a queryable tag recording that a resource employed a procedure — not the enacting of that procedure with participants and temporal unfolding (which is event-centered modeling). A description-of-method must not be mistaken for the methodology it merely points at.
Not useful by its mere presence. A populated method field carries retrieval weight only if its controlled vocabulary cuts finely enough and its tagging is applied consistently. A flat "survey" tag collapsing telephone, web, and randomized address-frame surveys into one bucket defeats the technique-filter it exists to provide; the axis's worth is set by granularity and discipline, not by existence.
Not a distinct cross-domain structure. The substrate-spanning impulse — expose a structural feature of a resource as a first-class, orthogonal, searchable axis — is carried by classification, index, and search_and_retrieval / paradigmatic_vs_syntagmatic_relations. Method-as-metadata is one named axis in that family; stripped of the cataloging framing the equivalent move is just tagging, too generic to be its own prime, so the abstract force lives upstream and the LIS machinery stays home.

Scope of Application¶

Because method-as-metadata is a descriptive-cataloging convention rather than a mechanism, it appears wherever a catalog serves users whose retrieval need is technique-compatibility and the procedure that produced each resource is recorded as a first-class, queryable field; the settings below are real uses of the same "method" axis across cataloging substrates, not analogies — the boundary is where that precondition holds, beyond which the move degrades into generic tagging.

Learning-resource cataloging (Dublin Core). instructionalMethod records the process a resource is designed to support, so teachers can filter by pedagogy (inquiry-based, lecture, simulation) rather than topic alone.
Dataset catalogs and data cards. Kaggle, HuggingFace, Zenodo, and domain-specific registries expose collection methodology ("survey," "scrape," "experiment") as a search facet for technique-matched discovery.
Clinical-trial registries. ClinicalTrials.gov records study_design (RCT, observational, single-arm) as a controlled-vocabulary field, enabling systematic-review authors to retrieve and meta-analyze by procedure.
Educational resource repositories (OER / LRMI). Instructional-approach metadata classifies materials by teaching technique so the corpus is filterable on pedagogy as a distinct axis.
Document and governance records. Process-type tags ("public comment period," "delegated decision," "fast-track review") let auditors aggregate records by the decision procedure used.

Clarity¶

Treating method as metadata makes explicit a retrieval need that topic-centered cataloging cannot serve: the user whose requirement is technique-compatibility, not subject-match. A systematic-review author does not want every paper on a condition — they want the randomized controlled trials; a teacher does not want every resource on a topic — they want the inquiry-based ones. Naming "method" as its own descriptive axis, parallel to topic, creator, date, and format, separates what a resource is about from how it was produced or what procedure it employs, two things that free-text description silently fuses. The payoff is a sharper question the catalog can now answer directly: "which resources use this technique?" — filterable, aggregable, and crosswalkable across a corpus rather than discoverable only by reading each item.

The move also locates precisely where such a field can fail, which keeps its design legible. Because method is recorded at cataloging time against a controlled vocabulary, its usefulness turns on the granularity of that vocabulary and the discipline of tagging — a flat "survey" tag collapses telephone, web, and randomized address-frame surveys into one undifferentiated bucket, defeating the very technique-filter the field exists to provide, whereas a hierarchical vocabulary preserves the distinctions a methodologist needs. So the field's worth is not settled by its mere presence but by how finely its values cut and how consistently they are applied — the question a cataloger must ask before the axis can carry weight.

Manages Complexity¶

When method lives only in each resource's free-text description, a user whose need is technique-compatibility faces a per-item reading task that scales with the whole corpus: to assemble every randomized controlled trial on a condition, or every inquiry-based resource on a topic, the systematic-review author or teacher must open each candidate, read it, and judge its procedure, repeated across thousands of items, the relevant ones scattered and indistinguishable from the irrelevant until read. Method-as-metadata compresses that task to a single filter. By lifting procedure to a first-class field drawn from a controlled vocabulary, it turns "read the corpus to find the technique" into "constrain one axis to a value" — the catalog returns the technique-matching set directly, and the same axis supports aggregation and crosswalking that reading never could. The analyst stops tracking the resources' contents and tracks instead one field per record, reading off membership in a technique class the way topic or date is read off. The compression is bounded by exactly one design parameter the Clarity section already isolates: the granularity of the vocabulary. A flat "survey" tag collapses telephone, web, and randomized address-frame surveys into one bucket, so the filter under-cuts and the methodologist must still read within the bucket; a hierarchical vocabulary preserves the distinctions, so the filter resolves to the precise technique. The branch structure of the field's usefulness therefore reads off two checks — is the vocabulary fine enough to cut where the user needs, and is the tagging applied consistently — and from those the analyst can predict whether the axis carries retrieval weight or merely appears to. So an open-ended "which of these many resources use this technique?" reduces to a one-axis query whose resolving power is set by a single vocabulary-granularity parameter.

Abstract Reasoning¶

Method-as-metadata licenses a set of reasoning moves built on two structural facts: that how a resource was produced is recorded as a queryable axis orthogonal to what it is about, and that the axis's resolving power is set by a single design parameter — the granularity of the controlled vocabulary.

Diagnostic — read retrieval need as a technique-compatibility question, and infer when topic-cataloging will fail. The characteristic move is to recognize, from the shape of a user's need, that subject-match is the wrong axis. A systematic-review author who needs the randomized controlled trials on a condition, a teacher who wants only the inquiry-based materials on a topic, an auditor aggregating governance records by decision procedure — each has a need that topic, creator, date, and format cannot serve, because the relevant items are scattered through the topic results and indistinguishable from the irrelevant until read. The inference runs from "the requirement is technique-compatibility, not subject" to "a method axis is necessary," and predicts that without one the user faces a per-item reading task scaling with the whole corpus. So the diagnosis identifies which retrieval needs the existing axes leave unmet, and attributes the failure to a missing descriptive dimension rather than to insufficient topic tagging.

Diagnostic of field quality — predict whether the axis carries weight from two checks, not from its mere presence. The concept's sharpest move is to refuse to credit a method field on existence alone and instead read its usefulness off two parameters. First, vocabulary granularity: a flat "survey" tag collapses telephone, web, and randomized address-frame surveys into one undifferentiated bucket, so the filter under-cuts and the methodologist must still read within the bucket; a hierarchical vocabulary preserves the distinctions and resolves to the precise technique. Second, tagging discipline: values applied inconsistently at cataloging time leave the axis unreliable even when fine-grained. The inference runs from these two checks to a prediction about resolving power — "is the vocabulary fine enough to cut where the user needs, and is it applied consistently?" — so the analyst can forecast whether a given method field will carry retrieval weight or merely appear to, before relying on it. A filter that returns a heterogeneous set localizes the fault to vocabulary granularity; one that misses known-matching items localizes it to tagging discipline.

Interventionist — lift procedure to a first-class field, and predict the read-the-corpus task collapses to a one-axis query. The corrective move is to record method as its own controlled-vocabulary field parallel to topic, and to predict the operational consequence: "read the corpus to find the technique" becomes "constrain one axis to a value," with the catalog returning the technique-matching set directly and the same axis newly supporting aggregation and crosswalking that per-item reading could never provide. Where the field under-resolves, the matched intervention is to deepen the vocabulary (split the flat "survey" into its sub-techniques), predicting the filter now cuts at the methodologist's needed grain; where it is inconsistently applied, the intervention is tagging discipline at description time. Each lever names the parameter it moves, and the prediction is that retrieval power rises only to the resolution the vocabulary supplies.

Boundary-drawing — separate the method axis from topic, from the method itself, and from the process it describes. Several lines the concept draws. First, what a resource is about versus how it was produced: free-text description fuses these, and the move is to hold them as distinct axes so a technique-filter does not silently degrade into a topic-filter. Second, the method axis is not any particular method-value: "inquiry-based" or "randomized controlled trial" are values the field can take, distinct from the prime-level move of having a method axis at all — so the move is to reason about the dimension, not to confuse it with one of its entries. Third, and subtler, the field is a static descriptor about a process, not the process itself: method-as-metadata records that a resource employed a procedure without enacting any temporal unfolding, so the move is to scope the diagnosis to the queryable tag and not mistake a description-of-method for the methodology it names. Drawing these boundaries keeps the analyst querying a single orthogonal axis whose resolving power is governed by vocabulary granularity, rather than conflating method with topic, with its own values, or with the process it merely points at.

Knowledge Transfer¶

Method-as-metadata is a descriptive-cataloging convention — one named axis on a resource record — rather than a causal mechanism, and within the cataloging family it transfers as the same convention across substrates. The structural move (lift procedure to a first-class controlled-vocabulary field parallel to topic, creator, date, and form), the technique-compatibility diagnostic, the field-quality checks (vocabulary granularity and tagging discipline), and the operational payoff (read-the-corpus collapses to a one-axis query, plus aggregation and crosswalking) carry without translation across descriptive-cataloging settings: Dublin Core's instructionalMethod for learning resources, methodology facets on dataset catalogs (Zenodo, Kaggle, HuggingFace), study_design controlled vocabularies in clinical-trial registries (ClinicalTrials.gov), OER/LRMI instructional-approach fields, and process-type tags in document and governance records. The concrete craft lessons — use a controlled vocabulary for methods, design crosswalks between different registries' method vocabularies, keep method distinct from topic — are real and transfer across all of these, because they are the same convention applied to one substrate (a field on a resource record). So within cataloging the transfer is genuine.

Beyond cataloging the honest account is the third category: the substrate-independent structural force is carried not by "method-as-metadata" but by the more general primes it inherits from, and those genuinely recur. The underlying impulse — expose a structural feature of a resource as a first-class, orthogonal, searchable axis — is the same one behind paradigmatic_vs_syntagmatic_relations (choosing an orthogonal descriptive axis), classification (the controlled value list of methods), and index / search_and_retrieval (the query affordance). Method-as-metadata is simply one more named axis in that family — the "method" axis — and where the cross-domain lesson is wanted ("expressing a process-type as a first-class field changes what you can ask of a corpus") it should be carried by those parents, since the general move is "choose the right axes for description," not anything specific to method. Tellingly, stripped of the library-and-information-science framing, the equivalent move outside cataloging is just tagging, which is too generic to be a distinct prime — confirming that the abstract force lives upstream and the LIS-tinted machinery (DCMI Terms, instructionalMethod, study-design enumerations, crosswalk authority layers) stays home. A neighboring boundary reinforces the limit: the field is a static descriptor about a process, not the process itself — enacting the procedure with participants and temporal unfolding is event_centered_modeling, a separate move — so method-as-metadata must not be mistaken for the methodology it merely points at. The "method" axis is the domain-specific instance; the substrate-spanning content belongs to the indexing/classification primes it instantiates — exactly the split drawn in Structural Core vs. Domain Accent.

Examples¶

Canonical¶

ClinicalTrials.gov's Study Design record is the cleanest worked instance. Every registered trial carries a structured Study Design block drawn from controlled vocabularies: a top-level Study Type (Interventional vs. Observational), and for interventional trials a hierarchy of Allocation (Randomized / Non-Randomized), Intervention Model (Parallel, Crossover, Factorial), Masking (Open, Single, Double, Quadruple), and Primary Purpose (Treatment, Prevention, Diagnostic). A systematic-review author seeking randomized controlled trials of a drug sets Study Type = Interventional and Allocation = Randomized, and the registry returns exactly the RCTs — no need to open each protocol and read its methods paragraph. The same fields let the reviewer aggregate and count trials by design.

Mapped back: The Study Design block is the designated method field, holding how the trial was conducted apart from what condition it studies — the orthogonality to topic. Its Randomized/Non-Randomized and Masking enumerations are the controlled method vocabulary, and their nesting is the granularity parameter set fine enough that filtering resolves to the precise technique. Setting Allocation = Randomized is the technique-compatibility query, and getting the RCT set back directly is the corpus-to-query collapse.

Applied / In Practice¶

Learning-resource repositories deploy the same move through Dublin Core's instructionalMethod and the LRMI educationalUse / instructional-approach fields. A platform cataloging teaching materials tags each item not only with its subject and grade band but with the pedagogy it supports — "inquiry-based," "direct instruction," "simulation," "problem-based learning" — drawn from a controlled list. A teacher who wants only inquiry-based science activities filters on the method field and the topic field together, retrieving materials that match both how they want to teach and what they want to teach, rather than skimming every science resource to infer its pedagogy.

Mapped back: The instructionalMethod value is the per-resource tag on the designated method field, and the enumerated pedagogies are the controlled method vocabulary. Combining it with a subject filter exercises the technique-compatibility query while preserving the orthogonality to topic — pedagogy and subject stay separate axes. The field's usefulness rests on the same tagging-discipline check: catalogers must apply the pedagogy values consistently for the filter to return a trustworthy set.

Structural Tensions¶

T1: Orthogonality to topic versus method-topic entanglement (a clean axis the resources resist). The move's foundational commitment is that "how a resource was produced" is a separate axis from "what it is about," so a technique-filter never degrades into a topic-filter. But real resources violate the separation: a paper about randomized-trial methodology has RCT as both its topic and, arguably, its method; a dataset whose subject is survey techniques blurs the two; and in methods-heavy fields the procedure is much of what the resource is about. Forcing a clean method axis onto such hybrids either double-classifies them or picks one axis arbitrarily, and the orthogonality that makes the filter powerful is a design ideal the corpus does not always honor. The axis is cleanest exactly where method and subject are already distant, and muddiest where a methodologist most wants precision. Diagnostic: For this resource, is the procedure genuinely orthogonal to what it is about, or is the method part of its subject — so that separating the axes distorts rather than clarifies?

T2: Granularity versus tagging discipline (the two quality checks pull against each other). The field's resolving power rises with vocabulary granularity — a hierarchical scheme distinguishing telephone, web, and address-frame surveys where a flat "survey" tag collapses them. But the concept's other quality condition, consistent tagging, degrades as granularity rises: a deep hierarchy is harder to reach consensus on, offers more nodes to disagree about, and widens the gap between how a cataloger classifies and how a searcher expects to find it, so finer vocabularies are tagged less reliably and queried less accurately. The two levers the concept isolates are therefore not independent — pushing granularity up tends to push discipline down — and a field can under-cut for being too flat or misfire for being too fine to apply consistently. Maximizing one check can spoil the other. Diagnostic: Is the vocabulary fine enough to cut where users need and coarse enough that catalogers apply it consistently, or has added granularity bought resolution at the cost of reliable tagging?

T3: The corpus-to-query collapse versus the unverifiable tag (retrieval by label, not by method). The payoff is that "read every item to find the technique" becomes a one-axis filter — but the filter returns items by their tag, not by their actual procedure, and the tag is an assertion made at cataloging time that the metadata cannot itself verify. A trial recorded as "randomized" that was not properly randomized, an "inquiry-based" resource so labeled for marketing, a self-reported method that misdescribes the process — all pass the filter. Worse, the very efficiency the field delivers (no need to open each item) removes the reading step that would have caught the mistagging. So the axis trades per-item verification for scale, and its trustworthiness rests entirely on a tagging fidelity it cannot check. Diagnostic: Does the method tag faithfully report what the resource actually did, or is the filter returning a set trusted on the strength of labels no one re-examines?

T4: Cataloging-time fixity versus evolving methodological categories (a frozen snapshot). Method values are assigned once, against a controlled vocabulary chosen at a point in time — but methodological practice keeps refining its categories: new techniques appear, old buckets split, distinctions the original vocabulary never anticipated become the ones reviewers now need. A legacy corpus tagged under yesterday's scheme cannot answer today's finer query without costly re-tagging, and crosswalks between different registries' vocabularies are lossy where their category boundaries disagree. So the axis that makes a corpus filterable also freezes it to the methodological ontology of its cataloging moment, and the field ages against a field of practice that does not hold still. Diagnostic: Does the vocabulary still reflect the distinctions users now need, or is the corpus locked to a superseded scheme that a current technique-query cannot resolve without re-cataloging?

T5: Autonomy versus reduction (a named LIS axis or the indexing primes — nearly just tagging). Method-as-metadata is a real, worked cataloging convention with its own machinery — DCMI's instructionalMethod, ClinicalTrials.gov study-design enumerations, crosswalk authority layers — and within cataloging it transfers intact as the same "method" axis. But its portable force is upstream, in classification (the controlled value list), index/search_and_retrieval (the query affordance), and paradigmatic_vs_syntagmatic_relations (choosing an orthogonal axis), and the concept nearly concedes its own non-autonomy: stripped of the library-science framing, exposing a resource feature as a first-class searchable axis is just tagging, too generic to be a distinct prime. A neighboring line reinforces the limit — the field is a static descriptor about a process, not the process enacted, which is event_centered_modeling. Diagnostic: Resolve toward the indexing/classification primes (and plain tagging) when the lesson is "choose the right descriptive axes"; toward the named method axis only when working inside a cataloging system's method vocabulary and crosswalks in situ.

Structural–Framed Character¶

Method-as-metadata sits on the framed-leaning stretch of the spectrum: a descriptive-cataloging convention, not a mechanism, wholly constituted by a documentary practice and authored by standards bodies — the entry nearly concedes its own non-autonomy, noting that stripped of the library-science framing the move is "just tagging." On evaluative_weight it is mostly neutral: a method axis is a descriptive dimension, not a verdict, and naming it convicts nothing — though the field-quality apparatus (vocabulary granularity, tagging discipline, whether the axis "carries weight or merely appears to") introduces normative judgments about the artifact, a mild framed streak. On human_practice_bound it is high: the convention exists only inside the practice of descriptive cataloging — a cataloger assigning a controlled-vocabulary value to a resource record so a retrieval system can filter — and it dissolves the moment that practice is removed; there is no observer-free "method-as-metadata," only the human act of description. Institutional_origin is strongly framed: the convention is an artifact of library-and-information-science standards traditions — Dublin Core's instructionalMethod, ClinicalTrials.gov's study_design enumerations, LRMI instructional-approach fields, DCMI Terms and crosswalk authority layers — drawn up by standards bodies rather than found in nature. Vocab_travels is low for the named layer (that LIS-tinted machinery stays home), and on import_vs_recognize the transfer within cataloging is re-recognition of the same convention across substrates (learning resources, dataset catalogs, trial registries, governance records), while the substrate-spanning force belongs upstream — and the equivalent move outside cataloging degrades into generic tagging.

The portable structural skeleton is expose a structural feature of a resource as a first-class, orthogonal, searchable axis. That impulse is genuinely substrate-spanning, but it is precisely what the "method" axis instantiates from its parents — classification (the controlled value list), index/search_and_retrieval (the query affordance), and paradigmatic_vs_syntagmatic_relations (the choice of an orthogonal descriptive axis) — not what makes "method-as-metadata" itself travel: the cross-domain reach ("choosing the right axes for description") belongs to those indexing/classification primes, while the LIS craft — the method vocabularies, the crosswalks, the study-design enumerations, and the static-descriptor discipline that holds a tag about a process apart from the process itself (which would be event_centered_modeling) — is the domain accent that stays home. Its character: a practice-constituted, standards-body-originated cataloging convention, structural only in the expose-a-searchable-axis skeleton it borrows from the indexing and classification primes — a skeleton so general that stripped of its library-science clothing it is barely more than tagging — which is exactly what leaves it framed-leaning rather than mixed.

Structural Core vs. Domain Accent¶

This section decides why method-as-metadata is a domain-specific abstraction and not a prime, and it carries the case for its domain-specificity — there is no separate section for that.

What is skeletal (could lift toward a cross-domain prime). Strip the cataloging framing and a thin relational structure survives: expose a structural feature of a resource as a first-class, orthogonal, searchable axis drawn from a controlled value list, so a corpus becomes filterable on that feature without inspecting each item. The pieces that travel are abstract — a descriptive axis held apart from the others, a controlled vocabulary supplying its values, a per-item assignment, and the retrieval payoff by which reading-the-corpus collapses to constraining-one-axis. That impulse is genuinely substrate-spanning, which is exactly why the catalog already houses it as the primes method-as-metadata instantiates: classification (the controlled value list), index and search_and_retrieval (the query affordance), and paradigmatic_vs_syntagmatic_relations (the choice of an orthogonal descriptive axis). But it is the core it shares, not what makes method-as-metadata distinctive — and the entry is unusually candid that stripped of its library-science clothing the move is barely more than generic tagging, too thin to be its own prime.

What is domain-bound. The content that makes it the method axis in particular is library-and-information-science furniture, and it does not survive extraction: the commitment to record how a resource was produced or what procedure it employs; the standards-body machinery (Dublin Core's instructionalMethod, ClinicalTrials.gov study_design enumerations, LRMI instructional-approach fields, DCMI Terms); the granularity and tagging-discipline quality checks that decide whether the axis carries weight; the crosswalk authority layers between registries' method vocabularies; and the static-descriptor discipline that holds a tag about a process apart from the process enacted (which would be event_centered_modeling). These are the worked vocabulary, the instruments, and the empirical cases the field actually studies. The decisive test: remove the cataloging practice and "record how it was made as a searchable field" is no longer method-as-metadata but plain tagging — there is no controlled method vocabulary, no crosswalk, no cataloging-time discipline, only an observer-free feature nobody is describing. The convention is constituted by the very documentary practice the prime bar asks it to shed.

Why this does not clear the prime bar. A prime is a relational structure whose vocabulary travels and whose cross-domain transfer is recognition of the same mechanism, not analogy. Method-as-metadata is an unusual case: it is not a mechanism at all, only a convention, and its reach is bimodal in a telling way. Within cataloging the convention transfers intact — the same "method" axis, the technique-compatibility diagnostic, the granularity-and-discipline checks, and the corpus-to-query collapse recur across learning-resource catalogs, dataset registries, trial registries, OER repositories, and governance records — but that is re-recognition of one convention under different vocabularies across a single substrate (a field on a resource record), not a mechanism spanning substrates. Beyond cataloging the move degenerates into generic tagging, too abstract to be proprietary. And when the substrate-independent lesson is wanted — "choose the right orthogonal axes for description, and a corpus becomes newly queryable" — it is already carried, in more general form, by the classification, index/search_and_retrieval, and paradigmatic_vs_syntagmatic_relations primes method-as-metadata instantiates. The cross-domain reach belongs to those parents; "method-as-metadata," as the named LIS axis, carries the standards-body craft that stays home.

Relationships to Other Abstractions¶

Current abstraction Method-as-Metadata Domain-specific

Parents (1) — more general patterns this builds on

Method-as-Metadata is a kind of Retrieval Facet Domain-specific

Method-as-Metadata is the production-procedure species of Retrieval Facet.

Children (1) — more specific cases that build on this

Faceted Vocabulary Domain-specific is part of, conditional Method-as-Metadata

Method-as-Metadata is an internal axis of a faceted vocabulary when procedure is exposed as a controlled, independently filterable field.

Not to Be Confused With¶

A topic / subject descriptor. The axis recording what a resource is about. Method-as-metadata records how it was produced or what procedure it employs — an orthogonal axis; a technique-filter ("find the RCTs") is a different query from a subject-filter ("find the studies on this condition"). Tell: is the field indexing the resource's aboutness (topic), or the procedure that made it (method)?
A particular method value. "Randomized controlled trial" or "inquiry-based" are values the field can take. Method-as-metadata is the move of having a method axis at all, not any single technique it names. Tell: is the referent one entry in the vocabulary (a method value), or the searchable dimension that admits many such entries (the method axis)?
The methodology itself (event-centered modeling). The enacting of a procedure with participants and temporal unfolding. Method-as-metadata is a static descriptor about that process — a queryable tag recording that a resource employed a procedure — not the procedure enacted, which is event_centered_modeling. Tell: is it a tag pointing at a procedure (method-as-metadata), or a model of the procedure running over time (event-centered modeling)?
Provenance metadata. The record of a resource's origin, lineage, and custody chain — who created or altered it and when. Method concerns the technique or procedure that produced it, not the ownership/transformation history. They overlap but answer different questions. Tell: is the field tracing where the resource came from and who touched it (provenance), or the method it employed (method-as-metadata)?
Free-text / uncontrolled tagging (folksonomy). Ad hoc keyword labels applied without a governed value list. Method-as-metadata draws from a controlled (often hierarchical) vocabulary applied with cataloging discipline; its resolving power depends on that granularity and consistency, which folksonomy lacks. Tell: are the values drawn from a governed controlled vocabulary with crosswalks (method-as-metadata), or freely coined keywords (folksonomy)?
classification / index / search_and_retrieval (the parents). The substrate-neutral primes — a controlled value list, a query affordance, the choice of an orthogonal descriptive axis — that carry the general lesson "expose a resource feature as a first-class searchable axis." Method-as-metadata is the LIS "method" instance; stripped of cataloging it is barely more than tagging, and the portable force lives with these parents. Tell: is the claim the general choose-the-right-axes move (the parents), or specifically the method vocabulary with its standards-body craft (method-as-metadata)? (Treated more fully in an earlier section.)

Neighborhood in Abstraction Space¶

Method-as-Metadata sits in a moderately populated region (41^st percentile for distinctiveness): it has near-neighbors but no dense thicket of look-alikes.

Family — Faceted Classification & Metadata (6 abstractions)

Nearest neighbors

Topic Facet — 0.89
Form Facet — 0.86
Theme Reification — 0.85
Geographic Facet — 0.84
Faceted Vocabulary — 0.83

Computed from structural-signature embeddings · 2026-07-12