006·318

data labeling

/ˈdeɪtə ˈleɪbəlɪŋ/ - n.

1 [colloq.] The step where the team discovers it never agreed on the categories.Keep. Punchy.This is the problem.

Working definition

2. The process of attaching agreed categories or values to examples so a model has something to learn from.

Evidence

See also

ground truthOne analyst's spreadsheet, promoted to truth because no one else volunteered.
inter-annotator agreementThe metric that quietly reports how undefined the term was all along.
label driftDefinition drift, now with a confidence interval.
training dataWhatever was easiest to export, now load-bearing.