A Multi-Axial Mindset for Ontology Design Lessons from Wikidata's Polyhierarchical Structure

A Multi-Axial Mindset for Ontology Design: Lessons from Wikidata’s Polyhierarchical Structure Ege Atacan Do˘gan, Peter F. Patel-Schneider Julius-Maximilians-Universit¨at W¨urzburg, egeatacandogan@gmail.com Independent Researcher, pfpschneider@gmail.com Abstract. Traditional ontology design emphasizes disjoint and exhaus- tive top-level distinctions such as continuant vs. occurrent, abstract vs. concrete, or type vs. instance. These distinctions are used to structure unified hierarchies where every entity is classified under a single upper- level category. Wikidata, by contrast, does not enforce a singular foun- dational taxonomy. Instead, it accommodates multiple classification axes simultaneously under the shared root class entity. This paper analyzes the structural implications of Wikidata’s polyhierarchical and multi-axial design. The Wikidata architecture enables a scalable and modular ap- proach to ontology construction, especially suited to collaborative and evolving knowledge graphs. Keywords: Wikidata · Ontology · Multi-Axial Classification · Polyhierarchy · Knowledge Graph Architecture 1 Introduction Ontology design has traditionally adhered to disjoint and exhaustive top-level distinctions, such as continuant/occurrent, or abstract/concrete. These distinc- tions are what we call the primary split of the ontology. Foundational on- tologies like BFO, DOLCE, and SUMO then extend this top-level split to a tree-shaped upper ontology grounded mostly in binary oppositions. Wikidata, by contrast, adopts a structurally different approach. Rather than committing to a single primary split, it organizes knowledge using multiple clas- sification axes simultaneously. The root class entity (Q35120) serves as the starting point for several conceptually distinct and sometimes overlapping trees, such as the abstract/concrete axis, the individual/collective axis, and the ob- servable/unobservable axis. These axes consist of mutually disjoint and complete (exhaustive) classes. However, a given class in one axis is not disjoint with a class in another axis. While having many direct subclasses under entity may seem 0 This paper was prepared with assistance from ChatGPT-4. The model supported drafting, structuring, and refining arguments. All interpretations and conclusions remain the responsibility of the authors. arXiv:2512.12260v1 [cs.AI] 13 Dec 2025 2 E. A. Do˘gan, Peter F. Patel-Schneider odd compared to the sparseness of other foundational ontologies, this can work well in Wikidata’s flexible structure. This paper presents a structural analysis of Wikidata’s ontology and intro- duces the concept of a multi-axial mindset for ontology design. We argue that the core difference lies in the architectural design: Wikidata accommodates mul- tiple orthogonal classification axes, permitting non-exclusive categorizations at all levels of the hierarchy. This allows entities and classes to participate in sev- eral high-level taxonomies simultaneously, in contrast to foundational ontologies that impose a single primary split. 2 Top-Level Ontological Distinctions A primary split in an ontology is the initial, high-level division that structures the class hierarchy in the ontology. It defines the most fundamental categories under which all other entities are subsumed. These splits are typically exhaustive and disjoint: every entity is expected to fall under exactly one branch, and no entity should belong to more than one. The choice of primary split reflects philosophical assumptions about the nature of reality, such as whether time, materiality, or identity is foundational. Logically, any distinction within an ontology can be a primary split, even a trivial one such as apples/non-apples, although the intuitive understanding of classification makes this approach non-sensical. Therefore, all primary splits should have a philosophical basis. Foundational ontologies such as BFO [1], DOLCE [8], SUMO [9], UFO [4]1, and Cyc [5] structure their class hierarchies according to disjoint and exhaustive top-level distinctions. These distinctions are meant to be the primary split of said ontology. Wikidata, by contrast, supports a flexible, multi-axial design in which such distinctions coexist as overlapping, non-exclusive axes under a shared root class: entity (Q35120). Table 1 compares key ontologies along their primary splits. Table 1. Comparison of Top-Level Ontological Distinctions Ontology Primary Split(s) BFO Continuant / Occurrent DOLCE Endurant / Perdurant / Quality / Abstract SUMO Physical / Abstract UFO Type / Individual Cyc Individual Object / Intangible / Represented Thing These ontologies are based on a classification such that every entity belongs to exactly one subclass at almost all splits. Classification thus generally proceeds 1 UFO has 3 other axes than given in the table, but Type/Individual is the one handled at the first split under “thing”. Multi-Axial Ontology Design in Wikidata 3 through mutually exclusive branche

This content is AI-processed based on open access ArXiv data.