We are exposed to a wide range of chatbots in our everyday activities—Siri, Alexa, and Google Home are just a few examples. I can speak out loud to various devices with utterances, such as, Siri, what is the weather in Seattle? or Alexa, Turn on the lights, or Google, What is the tallest mountain in the world? And these devices just respond with answers or actions.
What they do is fairly remarkable in that there are many ways of inquiring about the weather. For example, I might say, What is the weather in Washington?, Tell me the temperature in Seattle, or even the more ambiguous, What is it like up North? All of these have essentially the same intent and are generally referring to the same location. For the most part, all of these devices are doing the same thing: using machine learning algorithms to classify your intention, and to determine the entity (location, in this case) you are inquiring about. They then need to map that onto some predefined...