The input schema is the same used with the sample of suspect pairs generated
by the tMatchPairing component. The column LABEL
column holds the label you manually set on every second record in a
pair.
For readability purposes, you can ignore the columns PAIR_ID and SCORE.