How to train your tagger

Mihailo Škorić

In a setting where multiple automatic tagging and annotation technologies coexist and advance separately, but none of completely solves a specific problem, the key just might be in their cooperation.

This work is based on an old motto “Two heads are better than one” and explores an approach to tagging and annotation based on the incorporation of multiple tagging systems into a complex one, focusing on different types of their entanglement based on probability outputs of such.