A classification system is most effectively defined by concepts rather than words. A term may be too broad, it may carry several meanings, or it may not exist at all for a concept that matters. Each entry here is therefore treated as a concept. It is defined by a specific intention, disambiguated from its neighbors, and situated by facets rather than by spelling alone. That treatment lets every item land in the most specific concept possible, where other researchers can find it efficiently.
From the project thesis — Wong, The Need for a Classification System in Communication Design (project-held proposal, unpublished; see Overview → Thesis). The concept-over-term principle, the disambiguation, and the specific-concept aim are the proposal's; the facet apparatus is this system's own implementation of them.
Search and browse. Type in the search box to filter by term or definition text. The view toggle switches between grouping entries by Domain (18 top-level subject areas) and a straight alphabetical list. Four facets along the side — Domain, Kind, Medium, Scope — narrow the list further; check any combination of boxes, or click a colored chip directly on an entry to activate that same filter without opening the sidebar.
Chips feature a category. Every entry also shows its Domain, Kind, Medium, and Scope as small colored labels. Each one is clickable: selecting a chip clears whatever filters were active and features every term sharing that one value alone — a clean, standalone view of a whole category. The sidebar checkboxes work differently and still combine: check several boxes there to build up a specific intersection.
The Concept Map. Entries linked into the taxonomy carry a ◈ Concept map button. It opens a panel built from several kinds of relationship: Trails, the breadcrumb path (or paths) from this concept up to its root; Narrower, its children in the taxonomy; Involves / Involved by, a cross-cutting relationship distinct from parent–child, connecting concepts that co-occur in practice without one containing the other; and Related, a looser associative link. A word-family row, where relevant, groups the concept with other forms of the same root word (Design / Designing), since English regularly turns one concept into several parts of speech. Clicking any linked term inside the panel re-opens the map centered on that term, so the whole taxonomy is browsable by hopping from node to node.
Why a term can have more than one Trail. The taxonomy is a polyhierarchy — formally a directed acyclic graph (DAG) — not a strict tree: a concept may have more than one parent, so long as no chain of parents ever loops back on itself (verified across the full graph: zero cycles). 77+ concepts in this build have multiple parents. Six upper-ontology roots anchor every trail — Artifacts, Concepts, Practices, Events, Places, Eras; the Kind facet's eleven values are a finer classification assigned from each term's own position in the taxonomy, so several Kinds (Framework, Discipline, Role, Format, Metric) cut across more than one root rather than sitting inside a single one. A concept with two parents shows two Trails, each a legitimate, independently correct path to a root; neither is a duplicate or an error, and the map shows both rather than forcing a single canonical placement.
The structure, drawn. One real entry — Graphic Design — with every kind of edge the glossary supports:
Graphic Design carries two independent Trails — Concepts › Design › Communication Design › Graphic Design, and the shorter Concepts › Design › Graphic Design — because both Communication Design and Design are its parents. Across the 16th-edition build: 1,804 glossary terms are linked into a taxonomy of 1,627 concepts, held together by 1,706 parent edges (77 concepts have two or more parents; zero cycles), plus 154 involves and 31 related lateral edges. 177 entries carry a total of 183 cross-references to other volumes. Facet inventories: Domain 18 · Kind 11 · Media 7 (24 entries multi-valued) · Scope 2.
Diagram: editorial synthesis (GEN), rendered from the glossary’s own concept-map and facet data; all counts re-verified against the 16th-edition dataset, 22 Jul 2026 (unchanged by the root renames).
The Lineage line. Some entries carry a second provenance line, below the Source line, labeled Lineage. It records something the Source line does not: where an entry sits in a named theoretical tradition. The two are deliberately distinct — the Source records where this entry's wording came from (a course, a textbook, general knowledge), while the Lineage records whose framework the concept belongs to. A definition can be worded from a CityTech course handout while the concept itself descends from Peirce or from Gestalt psychology; the Lineage line names that intellectual parentage so the entry's wording source is not mistaken for the origin of the idea. Lineage notes carry the same three provenance tiers as any other claim in this system: SRC (source-verified) where the attribution has been checked against the cited source, CAN (canonical) where it is a standard scholarly ascription, and GEN (general) where it is editorial inference. Where a tradition has its own entry in the Theories document, the Lineage line points to it, so the concept's framework and that framework's own sources are one hop away. Four entries carry a Lineage line in this build: Icon / Iconography (Peircean semiotics); Symbols (Peircean semiotics); Figure/Ground Relationship (Gestalt, via the ADV 1100 departmental vocabulary); and Wicked Problems (Rittel & Webber 1973, via Buchanan 1992) — as the initial application of the convention; it extends outward from the entries that already cross-reference the Theories volume.
A classification is a set of commitments, and commitments invite the question: why these? The choices below carry the structure, and each had live alternatives.
What the six roots actually are. Read the singular field in the source data and the ontological character becomes obvious: physical thing / concept / activity / event / location / era. That is very close to the fundamental partition used by every general upper ontology: material entities, abstract entities, processes, occurrences, spatial regions, temporal regions. BFO splits continuant from occurrent, partitions independent continuants into material and immaterial entities — with spatial regions among the immaterial continuants — and places temporal regions on the occurrent side (Arp, Smith & Spear 2015; BFO is also ISO/IEC 21838-2:2020). DOLCE splits endurant, perdurant, quality, and abstract (Gangemi et al. 2002). Schema.org makes the same cut, less formally: Thing is its single root, with Action, CreativeWork, Event, Intangible, Organization, Person, Place, and Product among Thing's direct subtypes (schema.org, verified 22 Jul 2026). CIDOC CRM — the closest institutional cousin, since it is the cultural-heritage community's answer to this same problem — uses E70 Thing, E77 Persistent Item, E5 Event, E53 Place, E52 Time-Span (CIDOC CRM v7.1.3). These six roots are not idiosyncratic; they are a local instance of a partition that has been argued for, in exactly this form, for thirty years.
Why the partition is the right structural choice for this system. A partition — where every concept sits under exactly one root by kind-of-thing — is what lets Kind be a facet rather than a competing hierarchy. The moment you promote Framework, Discipline, Role, Format, and Metric to roots, you cross-classify: Graphic Design is a Discipline and an abstract Concept; Design Thinking is a Framework and a Practice. Every concept then belongs at two root positions for two different reasons, and the polyhierarchy stops being a genuine DAG in a meaningful sense — it becomes a lattice with mixed semantics. Keeping ontological type at the roots and functional type at the facet layer is standard faceted-classification practice (Ranganathan; Bruno & Richmond 2003) and it is what makes the retrieval structure work.
What each root holds — scope notes.
Artifacts. Holds material entities: tools and infrastructure of production (software, presses, cameras), media substrates (paper, screens, signage), and produced objects (books, posters, packaging, artifacts of design practice). This is the material-entity root: BFO's material entity, a subclass of independent continuant (Arp, Smith & Spear 2015); DOLCE's physical endurant (Gangemi et al. 2002); CIDOC CRM's E22 Human-Made Object under E70 Thing (v7.1.3). Renamed from "Tools" in the 16th edition because the root holds all material entities, not only instruments of production — the source data's own singular field reads physical thing. A typeface may sit under Artifacts as font software and also under Concepts as a design form; a poster sits here as a produced object with its format recorded under Concepts and its Medium recorded as a facet — the polyhierarchy carries both readings when both are meant.
Concepts. Holds abstract entities: ideas, theories, frameworks, principles, styles as ideas, and disciplines. This is the immaterial-entity root, corresponding to BFO's generically dependent continuant (Arp, Smith & Spear 2015), DOLCE's abstract (Gangemi et al. 2002), Schema.org's Intangible — a direct subtype of Thing — and CIDOC CRM's E28 Conceptual Object (v7.1.3). It is the largest root — 62 direct children in the current build — because most of what a research classification tracks (theoretical positions, methods as ideas, disciplines as fields of study) is abstract in this sense. Graphic Design sits here as a discipline; the activity of designing sits under Practices. Bauhaus sits under three roots at once — as a movement-Concept, a school-Place, and a period-Era — each capturing a distinct sense the corpus records.
Practices. Holds activities and processes at the type level: design methods, workflows, forms of practice, teaching, critique, publishing, curation. Corresponds to DOLCE's perdurant (Gangemi et al. 2002), BFO's occurrent — specifically process (Arp, Smith & Spear 2015) — Schema.org's Action, and CIDOC CRM's E7 Activity (v7.1.3). Renamed from "Method" in the 16th edition because the root holds activities broadly, not the narrower category of methods — the source data's singular field reads activity. The distinction between Practices and Events departs from CIDOC CRM, where E7 Activity is a subclass of E5 Event: here, Practices holds patterns of activity (Design Thinking, ethnographic fieldwork, the crit), and Events holds bounded historical occurrences. Design Thinking is a Practice; the launching of the Stanford d.school is an Event. A longitudinal case study can appear in both — as a practice type and as a specific project — and the polyhierarchy carries both readings when both are meant.
Events (repluralized, 16th ed.). Holds bounded historical occurrences: exhibitions, conferences, projects, launches, specific incidents. Corresponds to CIDOC CRM's E5 Event (v7.1.3), BFO's occurrent (Arp, Smith & Spear 2015), DOLCE's perdurant (Gangemi et al. 2002), and Schema.org's Event. The plural form was adopted for consistency with Artifacts, Concepts, Practices, and Eras; the singular field remains event per the source data. Distinguished from Practices by boundedness — a specific occurrence versus a type of activity — and from Eras by extent, since an Event happens within a period rather than being one. Boundary case: a design movement is recorded across Concepts (the ideas), Eras (the period), and Events (the landmark exhibitions and publications that anchored it); Bauhaus is documented across all three.
Places (repluralized, 16th ed.). Holds spatial locations and place-associated institutions: cities, regions, schools, publishers, studios. Corresponds to CIDOC CRM's E53 Place (v7.1.3), BFO's spatial region — an immaterial entity on the continuant side (Arp, Smith & Spear 2015) — and Schema.org's Place. In strict upper ontology, institutions are actors or groups rather than places — CIDOC CRM separates E53 Place from E39 Actor and its subclasses (v7.1.3). This taxonomy collapses the distinction because design institutions are typically identified with their location: the Bauhaus is inseparable from Weimar/Dessau/Berlin, MIT Media Lab from its siting. The concession is deliberate and worth stating: where an institution is genuinely place-independent (a distributed research network, an international society), it sits under Concepts as an organization-idea instead, so the fudge does not propagate to cases where it fails.
Eras. Holds temporal regions: historical periods, epochs, movements as periods. Corresponds to CIDOC CRM's E52 Time-Span (v7.1.3) and BFO's temporal region — an occurrent (Arp, Smith & Spear 2015). Carries the source data's original description, a period in time — the only root that was ever defined in the source data, and the description is preserved as inherited. Distinguished from Events by extent: an Era is a period during which many events occur (Modernism, the digital revolution); an Event is a bounded specific occurrence within or across an Era. Boundary case: a design movement is often both a set of ideas (Concepts) and a period during which those ideas were dominant (Eras); the polyhierarchy carries both — the Bauhaus movement as a Concepts entry (form-follows-function, craft–industry integration); the Bauhaus period as an Eras entry (roughly 1919–1933).
Why a polyhierarchy and not a tree. Because the corpus is genuinely polyhierarchic: 77+ concepts have more than one legitimate parent (Graphic Design belongs to both Design and Communication Design), and a strict tree would force a single canonical placement the sources do not support. The cost — multiple Trails per concept — is accepted and displayed rather than hidden; the constraint kept is acyclicity, verified across the full graph.
Why six upper roots. The six roots — Artifacts, Concepts, Practices, Events, Places, Eras — partition: they answer different questions about a concept (what is it materially? what does it do? when and where does it sit?), and no entry needs two of them for the same reason. Whether the roots should nonetheless be restructured to align with the Kind facet's eleven values is an open architectural question, held open on the record rather than resolved quietly — it sits precisely on the line between this system and a formal ontology (see the self-classification note on the Overview). As of the 16th edition (22 Jul 2026), the roots carry ontological categories and the Kind facet carries functional types, so the two vocabularies now diverge by name as well as by role.
Why these four facets. Domain, Kind, Medium, and Scope follow the facets-as-metadata recommendation (Bruno & Richmond 2003): facets describe a concept from independent angles rather than nesting it deeper. Domain places a concept in the field's subject geography; Kind types its function; Medium records its channels (zero to many); Scope marks core discipline versus adjacent territory. Each facet is orthogonal to the trail structure — filtering by any combination never contradicts a concept's placement — which is what lets the thesis's placement rule (every item in its most specific concept) coexist with multi-angle retrieval.
Why not classification by medium. Because media are properties of concepts, not containers for them: the same concept crosses print and screen, and a medium-rooted scheme would split single concepts into parallel duplicates. Medium is therefore a facet here, never a root.
Why not adopt an existing scheme wholesale. None of the adjacent systems' units match this system's unit — the communication-design research concept. Object-cataloging vocabularies classify artifacts; subject headings classify publications; discipline-owned schemes classify other disciplines. The gap is specific enough to warrant a purpose-built system rather than a borrowed one.
Source line. Upper-ontology grounding per Arp, Smith & Spear (2015), MIT Press (BFO; also ISO/IEC 21838-2:2020); Gangemi, Guarino, Masolo, Oltramari & Schneider (2002), EKAW 2002, LNCS 2473, 166–181, doi:10.1007/3-540-45810-7_18 (DOLCE); CIDOC CRM v7.1.3 (cidoc-crm.org); Schema.org type hierarchy (schema.org/docs/schemas.html) — all SRC, live-verified 22 Jul 2026. Structural reasoning is editorial (GEN); partition and facet-vs-root reasoning per Ranganathan and Bruno & Richmond (2003); placement rule per the project thesis. Counts (77+ multi-parent; six roots; facet inventories) verified against the 16th-edition dataset, 22 Jul 2026. The six-vs-eleven roots question is logged as open in the changelog; the position defended here is the ontological grounding of the current partition, not a resolution of the roots-vs-Kind alignment question.
16th edition — upper-ontology roots renamed on ontological grounds (22 Jul 2026). Four of the six root display terms changed: Tools → Artifacts and Method → Practices (both original names were narrower than the ontological categories they hold — the source data's own singular fields read physical thing and activity), and Event → Events, Place → Places (repluralized for consistency). Concepts and Eras are unchanged. The renames follow an upper-ontology review grounding the six roots in the partition shared by BFO (Arp, Smith & Spear 2015; ISO/IEC 21838-2:2020), DOLCE (Gangemi et al. 2002), CIDOC CRM (v7.1.3), and Schema.org — documented, with verification, in the project record. Deliberately unchanged: the Kind facet's eleven values (Tool, Method, etc. remain as functional types — the roots now name ontological categories, so the two vocabularies diverge by design); the two pre-existing concepts sharing old root names (Tools, id 171, a child of Artifacts; Method, id 581, a child of Concepts); and, per the standing rename-fossil convention (7th ed.), every singular field and slug. Graph counts re-verified after the rename, 22 Jul 2026: 1,627 concepts, 1,706 parent edges, 77 multi-parent, 154 involves + 31 related lateral edges, zero cycles — unchanged by the rename, as a display-term change should leave them.
15th edition — No Film School, PetaPixel, B&H eXplora, and Sweetwater verified as sources (20 Jul 2026). Four production-craft glossaries live-verified against the Photography, Video & Audio domain: No Film School (nofilmschool.com, five glossary pages spanning filmmaking, editing, film theory, cinematography, and production-assistant vocabulary), PetaPixel (petapixel.com/photography-terms-glossary), B&H eXplora (bhphotovideo.com, digital photography terms), and Sweetwater’s inSync glossary (sweetwater.com/insync). 24 terms upgraded from general-knowledge-only (or single-sourced) to multiply-sourced: Rule of Thirds, Depth of Field, Foley, Sound Design, Non-Linear Editing (NLE), Compositing, Aspect Ratio, Voiceover, White Balance, Photogram, Product Photography, Histogram, Continuity Editing, Cinematography, Montage / Jump Cut, Color Grading vs. Color Correction, Coverage & Shot List, Frame Rate & Resolution, Codecs & Containers, Lens: Prime / Zoom & Focal Length, Exposure Triangle (Aperture / Shutter / ISO), Chroma Key / Green Screen, Photography (Commercial & Editorial), and Mixing & Mastering. No new terms added; compound terms were verified and cited as-is rather than split into atomic entries, preserving their comparative definitions (see README for the architectural rationale). Three partial matches, noted in their definitions: Chroma Key / Green Screen (No Film School verifies “Green Screen” specifically, not “Chroma Key”), Photography (Commercial & Editorial) (PetaPixel verifies the general term, not the application scoping), and Mixing & Mastering (Sweetwater verifies “Mixing” specifically; “Mastering” has no dedicated glossary headword there despite extensive article coverage). 1,802 terms.
14th edition — Adweek verified as a source (19 Jul 2026). Glossary of Marketing Terms (Adweek, adweek.com/glossary, retrieved 19 Jul 2026) live-verified as a reference glossary. Matched by exact title against the Advertising & Persuasion domain: seven terms upgraded from general-knowledge-only to sourced — Brand Awareness / Brand Recall, Brand Safety, Influencer Marketing, Native Advertising, Programmatic Advertising, Public Relations (PR), and Retargeting. No new terms added; this pass verifies existing entries only. Brand Awareness / Brand Recall is a partial match, noted in its definition — Adweek verifies “Brand Awareness” specifically, not the compound term’s “Brand Recall” half. 1,802 terms.
13th edition — Collection and Provenance filters removed (17 Jul 2026). Both facets, and the per-entry chips for them, are gone from the sidebar and from every entry; Domain, Kind, Medium, and Scope remain. Underlying col/p data kept on every entry — this removed the filter, not the provenance record. 1,802 terms.
12th edition — usability & error-review pass (17 Jul 2026). Full structural review (zero duplicate terms, two fragmented Provenance citations merged), Method & provenance pared to this 1–3-sentence-per-edition format, a “How to use this glossary” guide added, and every entry’s facet chips made clickable filters. 1,802 terms.
11th edition — full-site bidirectional coverage (17 Jul 2026). Extended full bidirectional linking to Theories, both Pedagogy Atlas layers, and the Crossover Map, and removed the site's remaining new-tab links. 1,802 terms.
10th edition — Research Processes methods & approaches (17 Jul 2026). Gave all 39 items in the Research Processes atlas's methods rails a glossary definition and a same-window link. 1,699 terms.
9th edition — cross-volume consolidation (16 Jul 2026). Reviewed every glossary-bearing volume site-wide: merged the Field Atlas's 43 unique terms, retired the duplicate Intersections vocabulary in favor of linking, and introduced the xrefs cross-reference mechanism for Theories and Research Processes. 1,673 terms.
8th edition — full concept-node coverage (16 Jul 2026). Added definitions for the 128 taxonomy nodes that had structure but no glossary entry, giving every one of 1,627 concepts exactly one definition. 1,630 terms.
7th edition — full taxonomy unification (16 Jul 2026). Restructured the glossary around the tag_concept polyhierarchy (638 concepts extended to 1,627 nodes), unifying Kind into a single 11-value classification and adding the Concept Map's breadcrumb navigation. 1,502 terms.
6th edition — anonymization & original-wording pass (15 Jul 2026). The course-list source was relabeled "ADV 1100 departmental vocabulary" for accuracy, and the 54 flagged entries were rewritten in original wording, closing the verification queue.
5th edition — named correction (13 Jul 2026). A plagiarism review traced the ADV 1100 vocabulary to uncredited textbook sources (Ocvirk et al., Stewart). Four entries were paraphrased with lineage noted; 54 more were flagged "under verification" pending resolution.
Merge correction (14 Jul 2026). "Visual Analysis" merged into "Visual / Formal Analysis" per the standing merge rule. 938 terms.
4th edition. Added the nine-term Critical Frameworks Supplement, live-verified. 939 terms.
3rd edition. Merged the Consolidated Glossary (771 entries) with the Design Incubation keyword corpus (101 entries) and 58 ADV 1100 terms. 930 terms.
Strip-down guarantee. Removing all supplement-sourced and general-knowledge material leaves only entries traceable to a named document, course, or publication.
Copyright © 2026 Dan Wong, New York, NY.