Essays at the seam

A concept lattice is, by construction, an interpretable artefact. What does it mean to use one as a model — and where does that view help, and where does it strain?

Galois connections are the algebraic skeleton inside Formal Concept Analysis. They are also, perhaps less obviously, the algebraic skeleton inside a number of learning algorithms — closure operators in disguise.

Preferences are partial orders, not real-valued scores. Most modern ML pipelines collapse them into scalars and lose much of what the labellers actually said. This essay works through the order-theoretic alternatives in detail.