Usually researchers manually code their selections and then rebuild the classifier, but there is the option to perform manual training in a code set.
This is an optional way of manually training a classifier if you already have some text (one training item per line in that text box) that you want to train the classifier with. This comes in handy if you already have some seed text for the class of the classifier that you know to be true. Using this dialog will basically just add more training data to a classifier's class, as if the text input was part of the dataset itself.
Most researchers never use this method, as the train-via-coding in DiscoverText is the preferred way of organically creating and training classifiers; however, the option is available.
To manually train a classifier:
- On the Dataset Details page, click on Advanced Options.
- Click on your Classifier name.
- On the right side of the page, click on Manual train.