constantinos orphanides.

research /

Turning structured data into formal concepts.

My research field is Formal Concept Analysis: a mathematical way of discovering which objects share which attributes, and of ordering those groupings into a hierarchy called a concept lattice.

The field belongs to the symbolic tradition of AI. Every concept in the lattice is derived from the data, and nothing appears in the output that the input does not support. That guarantee is unfashionable, and generative AI cannot offer one.

crypto 24h US-regulated traded
xmr × × · ×
es · × × ×
ibit × · × ×
btc × × × ×
Eight concepts. Top: traded, shared by btc, es, ibit and xmr. Next: 24h (btc, es, xmr); crypto (btc, ibit, xmr); US-regulated (btc, es, ibit). Next: crypto plus 24h (btc, xmr); 24h plus US-regulated (btc, es); crypto plus US-regulated (btc, ibit). Bottom: all four attributes (btc). Covering edges connect comparable concepts on adjacent levels. traded 24h crypto US-regulated xmr es ibit btc
a formal context and its concept lattice · hover over a node · click on a cell

thesis /

Appropriating Data from Structured Sources for Formal Concept Analysis 

PhD thesis · Sheffield Hallam University · 2022 · DOI 10.7190/shu-thesis-00503

publications /

profiles /