Data Annotation: "labeling" target vs features

Question

I understand how one would use a data annotation tool to label targets for a given sentence, for example though, I'm not clear on how placing labels on features can be used to improve model performance. For example, in this text annotation tool , you can add "labels" to a body of text like person, location, event ...etc . Given that you must create Word Embeddings to work with the data, and the vector representation is not human-readable, how would you be able to improve model performance by annotating feature variables?

Answer 1

Those labels are not primarily for features, those labels are primarily for targets. Person, location, and event for targets for named-entity recognition (NER).