Before we continue to build our classifier, let's imagine what the main function will look as follows. It will look something similar to this:

unc main() {
  examples, err := ingest("bare")
  log.Printf("Examples loaded: %d, Errors: %v", len(examples), err)
  shuffle(examples)

  if len(examples) == 0 {
    log.Fatal("Cannot proceed: no training examples")
  }

  // create new classifier
  c := New()

  // train new classifier
  c.Train(examples)

  // predict
  predicted := c.Predict(aDocument)
  fmt.Printf("Predicted %v", predicted)
}

The use of Train and Predict as exported methods are useful in guiding us on what to build next. From the sketch in the preceding code block, we need a Classifier type, that has Train and Predict at the very least. So we'll start by doing that:

type Classifier {}

func (c *Classifier) Train(examples []Example) {}

func (c *Classifier) Predict(document []string) Class { ... }

So, now, it becomes a question of how the classifier works.