Reviewing label predictions

User permissions required: 'View Sources' AND 'Review and annotate'.

Annotating overview

After the Discover phase, the model will start making predictions for many of the labels in your taxonomy.

The purpose of Explore phase is to review predictions for each label, confirming whether they are correct and correcting them where they aren't, and thereby providing many more training examples for the model.

There are two key actions when reviewing label predictions:

• Where the predictions are correct, you should confirm/accept them by simply clicking on them

• Where they are incorrect, you should either dismiss/ignore them or alternatively add the correct label(s) that does apply. To add a different label, click the ‘+’ button and type it in. This is the way to correct wrong predictions, by adding the correct one and not clicking on any incorrectly predicted labels

The images below show how predictions look in Communications Mining for data with and without sentiment. Hovering your mouse over the label will also show the confidence the model has that the specific label applies.

The transparency of the predicted label provides a visual indicator of the model's confidence. The darker the colour, the higher the confidence, and vice versa:

Message with predictions without sentiment enabled

Message with predictions with sentiment enabled

To delete a label you applied in error you can hover over it and an ‘X’ will appear. Click this to remove the label.

Important: Remember: Add ALL labels that apply to the message you're reviewing. Telling the model that a label does not apply is just as important as telling it what does.