Summary
In this chapter, we learned how to train an EntityLinker component using spaCy. We saw some implementation details to learn more about the HashEmbedCNN.v2 layer and the EntityLinker.v2 architecture.
We also discussed some characteristics of high-quality datasets for NLP training, stressing the importance of consistency, representativeness, and thorough documentation. Finally, we saw how to create a custom corpus reader used to train the entity linking model. With this knowledge, you can customize any other spaCy component.
In the next and final chapter, you will learn how to combine spaCy with other cool open source libraries to create great NLP applications. See you there!