Summary
In this chapter, you learned how to generate a complete semantic parse of utterances. First, you added a SpanRuler component to extract an NER entity that is significant to the use case context. Then, you learned how to use DependencyMatcher to perform intent recognition by analyzing sentence structure. Next, you also learned how to create your own custom spaCy component to extract the intent of the utterances. Finally, you saw how to process large datasets faster with the Language.pipe() method.
Both SpanRuler and DependencyMatcher rely on the patterns we create. The process of creating these patterns is a back-and-forth process. We analyze the results, then test out new patterns, then analyze the results again, and so on. The goal of this chapter was to teach you how to use these tools so you can perform this process in your own projects.
In the next chapters, we will shift more toward machine learning methods. Chapter 6 will cover how to use spaCy with Transformers...
