Methodology

An honest account of how this artifact was constructed, what decisions were made, and where it falls short. Read this before citing or critiquing the ontology proper.

Source text

The ontology is extracted from the Ethics in the Elwes translation (1883), which is in the public domain and available through Project Gutenberg. The Curley translation (Princeton, 1985 and 2016) is more accurate by current scholarly standards but is in copyright; using it as the source for a public artifact would have required permissions the project did not pursue. Where the Elwes rendering is unclear or misleading by current standards, the issue is noted in the relevant class's comment. The reader should cross-reference with a current edition for serious scholarly use.

Latin alignment is not currently included. The ontology is in English. A future revision may add a Latin parallel based on the Bruder edition (1843), itself public domain.

Pipeline

The ontology was constructed using BFO-Agent, a dialogue architecture that combines large language model extraction with reasoner-based validation. The architecture is documented separately; for the present purposes the relevant components are:

Extraction

Each chunk of source text is presented to a language model (Anthropic Claude, Sonnet variant) along with the existing state of the ontology. The model proposes new classes, individuals, and axioms keyed to BFO categories. Proposed commitments include source quotes from the chunk, which serve as provenance.

Validation

Each proposal is staged and tested against the existing ontology by the HermiT description-logic reasoner (Glimm et al., 2014). Proposals that produce inconsistency, or that violate BFO's standard disjointness axioms (continuant ⊥ occurrent, material ⊥ immaterial, role ⊥ disposition, etc.), are rejected automatically.

Review

Each consistent proposal is reviewed by the author. Acceptance is not automatic. Proposals that are formally consistent but philosophically unfaithful to Spinoza's text (interpretive errors, mismappings, or missed nuances) are rejected at this stage. The acceptance rate across the project is reported in the methodology paper.

Commit

Accepted proposals are merged into the working ontology and committed to git, with the source quote as commit metadata. The commit history is the audit trail; any state can be reproduced or rolled back.

Finalization

The ontology is finalized as a frozen artifact when extraction is complete. Finalization marks the ontology as read-only and stamps the manifest with date and DOI. Changes after finalization require a new version number.

Decisions worth flagging

Substance as independent_continuant

Spinoza's substance is rendered as a subclass of BFO's independent_continuant (BFO_0000004). This commits us to the view that substance, in Spinoza's sense, persists through time and is not dependent for its existence on any other entity. The mapping is uncontroversial. The harder choice was whether to type substance as a single named individual (Spinoza thinks there is one and only one substance) or as a class with one named instance (Deus). The latter was chosen for compatibility with extension to other ontologies that may treat substance differently.

Attribute as quality

Attributes are rendered as BFO qualities (BFO_0000019), inhering in the bearer Substance. This is the natural BFO mapping but it does not perfectly capture Spinoza's view that attributes constitute the essence of substance rather than merely qualifying it. The ontology marks the inheres-in relation but adds an additional class, AttributeAsConstitutiveOfEssence, to distinguish constitutive from accidental qualities.

Mode as specifically_dependent_continuant

Modes are rendered as BFO specifically dependent continuants (BFO_0000020) inhering in the substance whose attribute they modify. Finite modes and infinite modes are subclasses. This mapping captures Spinoza's claim that modes exist in and are conceived through something other than themselves. The dependence relation specifically_depends_on is invoked.

Conatus as disposition

The conatus (the striving by which each thing endeavours to persist in its being) is rendered as a BFO disposition (BFO_0000016). Dispositions are realizable entities whose realization is a process; conatus is realized in the actual striving of the thing. Alternative renderings (as a quality, as a process) were considered. The disposition mapping was chosen for its alignment with Spinoza's claim that conatus is the actual essence of the thing (Ethics III, Prop. 7), which suggests an intrinsic readiness to act rather than an actual action.

Affects as transitions

The 50+ affects defined in Part III are rendered as BFO processes (BFO_0000015), with the participants typed as the bearer (a finite mode) and the relevant ideas. This was the most extensive set of decisions, and the methodology paper discusses the alternatives in detail.

Limitations

Translation drift. The Elwes translation is a century older than current scholarly conventions; some of its terminology is now considered misleading. The ontology inherits these limitations.

Single rendering. The ontology represents one reading of Spinoza, by one extractor, validated by one reviewer. Different scholars would produce different ontologies. The artifact is offered as a starting point for that disagreement rather than as a definitive structure.

Reasoner caveats. HermiT validates consistency under standard OWL 2 DL semantics. Spinoza's metaphysics involves modal claims (necessity, eternity) and relational structures (parallelism between attributes) that OWL 2 DL renders only approximately. Modal logic extensions and second-order constructs are not used.

Proposer biases. The language model used for extraction has its own prior associations with Spinoza's text. Where the model misreads a passage, the misreading propagates to a proposal, which can pass HermiT (because it is locally consistent) and yet be wrong about Spinoza. Author review catches most such errors but not all. The provenance trail makes systematic correction possible.

Comment-as-resource bug. A bug in the proposer's commit pipeline occasionally emitted comment annotations as malformed XML. Approximately one in 6,000 commits produced this artifact; affected lines were corrected manually. The bug is documented in the BFO-Agent issue tracker.

References

Arp, R., Smith, B., and Spear, A. D. (2015). Building Ontologies with Basic Formal Ontology. MIT Press.

Glimm, B., Horrocks, I., Motik, B., Stoilos, G., and Wang, Z. (2014). HermiT: An OWL 2 reasoner. Journal of Automated Reasoning 53(3), 245–269.

Smith, B. and Ceusters, W. (2010). Ontological realism: A methodology for coordinated evolution of scientific ontologies. Applied Ontology 5(3-4), 139–188.

Spinoza, B. (1883). Ethics, trans. R.H.M. Elwes. London: George Bell and Sons. Public domain.