What is NLP?
Natural Language Processing (NLP) is part of the software that makes Alexa, Google, Siri, Bixby and other intelligent virtual assistants speak "human." In a different post, I discussed some of the core constructs of NLP, specifically intents, utterances and slots. It occurred to me, though, that it would be worthwhile taking that up a level and cover the concept of a Natural Language Model, which is at the heart of every voice application.
Natural Language Processing and Virtual Assistants
When people talk to a virtual assistant, their request is converted from audio to digital text. Software designed to understand various ways humans speak – natural language processing software – analyzes it to trying to determine what the intent of the request. Just as a computer is pretty useless without having a set of programming instructions telling it what to do, a virtual assistant can’t do much without a natural language model to provide intent.
What is a natural language model?
A natural language model for a voice application is comprised of intents, utterances, and slots that give a contextualized view to the underlying machine learning algorithms so they can more clearly determine what the user is saying. As part of developing a voice application, a developer not only has to write backing code, but also provide a well-thought out natural language model that will control how easily the user will be able to interact with the app.
When a virtual assistant is not acting as a proxy to a voice application, it still has a much larger, more generalized natural language model it works with to determine what the user is asking it to do. When a user supplies an invocation name to a virtual assistant, e.g. “Alexa, open Whetstone Technologies” where the invocation name in this case is “Whetstone Technologies”, that is essentially a hint to the backing machine learning algorithm that the user is interested in interacting with the Voice Application called “Whetstone Technologies”. The virtual assistant then starts the voice application and passes the user’s input through that application’s own specific natural language model to provide a contextualized experience until either the application stops or the user abandons the interaction.
What does NLP mean in voice assistant technology?
The bottom line is, when creating a voice application, in addition to useful content and services, a well thought-out and designed natural language model is required to ensure that a user has a friction-free and engaging experience.
Have further questions on this sort of topic or voice applications in general? Feel free to reach out to Whetstone Technologies via our Contact page to schedule a free one-hour consult.