ManualLabel.scala

// Tabulator: checkout http://stackoverflow.com/questions/7539831/scala-draw-table-to-console

def propertyList(kvKRecord: KvKRecord): List[Any] = KvKRecord.unapply(kvKRecord).map(_.productIterator.toList).getOrElse(Nil)

val labeledList: ArrayBuffer[LabeledVector] = ArrayBuffer()

breakable {
  comparableDataset
    .sample(withReplacement = false, Config.sampleFactor)
    .collect()
    .foreach {
      case (left, right, vector) => {
        val table = propertyList(left) zip propertyList(right) map (x => List(x._1, x._2))

        println(Tabulator.format(table))
        println("Same? [y/n]")

        Try(scala.io.StdIn.readChar()) match {
          case Success('y') => labeledList.append(LabeledVector(left.dossierNummer, right.dossierNummer, vector, 1.0))
          case Success('n') => labeledList.append(LabeledVector(left.dossierNummer, right.dossierNummer, vector, 0.0))
          case _ => break
        }
      }
    }
}

spark
  .createDataFrame(labeledList.toList)
  .write.mode(SaveMode.Append)
  .parquet(path)