Session Ready
Exercise

Data-efficient entity recognition

Most systems for extracting entities from text are built to extract 'Universal' things like names, dates, and places. But you probably don't have enough training data for your bot to make these systems perform well!

In this exercise, you'll activate the MITIE entity recognizer inside Rasa to extract restaurants-related entities using a very small amount of training data. A dictionary args has already been defined for you, along with a training_data object.

Instructions
100 XP
  • Create a config by calling RasaNLUConfig() with a single argument cmdline_args with value {"pipeline": pipeline}.
  • Create a trainer and use it to create an interpreter, just as you did in the previous exercise.