nx_flavor_pairing.nx

nx_flavor_pairing.nx -- T4 of the TASTE-SCIENCE ladder: the FLAVOR NETWORK. Ahn, Ahnert, Bagrow & Barabasi (2011, "Flavor network and the principles of food pairing") showed that ingredient pairing can be computed from the aroma compounds two ingredients SHARE -- a data- driven affinity, not a chef's opinion. The baseline nx_food_science carries a HAND-LISTED affinity table (beef-oyster = 9, ...): finite, and silent on any pair nobody typed in. T4 computes affinity from each ingredient's compound set, so it scores ANY pair -- including a novel ingredient that has never been in a table. shared(A,B) = |compounds(A) INTERSECT compounds(B)| affinity(A,B) = Jaccard = 1000 * shared / |A UNION B| (milli) MEASURED EXCEED (gate-checked): a fruity pair that shares esters scores above a fruit+allium pair that shares none, identical sets score 1000, and a NOVEL ingredient gets a real affinity computed from its compounds. genealogy_id: ahn_2011_flavor_network + jaccard_1901_similarity