Alexa, What is Natural Language Processing?

I suppose you could ask Alexa, but I’m going to answer it for you anyway. Natural language processing (NLP) is a branch of artificial intelligence (AI) that helps computers understand, interpret, and manipulate human language. Though NLP has been around for quite some time, the technology is advancing rapidly thanks to increased interest in human-to-machine interaction and the availability of big data and complex algorithms.  Commonly lumped together under the umbrella term NLP, the natural language processing “engine” is actually made up of three components that work together to facilitate technology-human communications. 

The Natural Language Processing Engine 

  • Natural Language Understanding (NLU):  NLU refers to the “reading” of human communications by computers to understand context and intent. 
  • Natural Language Processing:  NLP converts the understanding of text/voice into structured data. 
  • Natural Language Generation (NLG)NLG is what happens when computers generate responses based on structured data.  

This entire feedback loop was created because, unlike humans who communicate in nuanced (written and verbal) languages like English, Japanese, or Korean, computers can only understand structured data. How are computers with data consisting of 0s and 1s supposed to pick up on the unstructured nuances of human communication? Take sarcasm, for example.  When a person says “Really, Sherlock?” as a human we understand the intent of this phrase and the emotions this elicits. Computers do not.  

NLP focuses on enabling computers to process a human-level understanding of language and intent. Have you ever asked Alexa a kooky question, like “Alexa, make me a sandwich”? Chances are you were met with a sarcastic comment back, in this case “Okay, you’re a sandwich. (Click here for a few of our favorite Alexa Q&A’s) 

Human-to-Technology Communication 

Let’s take a closer look at the Alexa interaction because it perfectly illustrates the NLP engine at work. Your device activated when it heard your voice, understood the intent of the comment (NLU), and provided feedback (NLP) in a lifelike, conversational English sentence (NLG). The complete interaction was made possible by the NLP engine. (Though it’s important to note that while these interactions do take place, they have been pre-programmed and we’re not at the point where Alexa can ad lib responses to unexpected questions). We see examples of NLP in lifelike chatbots like Amazon’s Lex, but also on websites and apps we interact with on a daily basis.  

An example of this is in cloud communications platform Twilio, which counts Airbnb, Uber, and Nordstrom as clients. Amazon Lex provides developers on Twilio apps a modular architecture with application programming interfaces (APIs) to enable building and deploying conversational bots on mobile platforms. Things like language translation, semantic understanding, and text summarization are being deployed (as a service) in most technology stacks.

What’s Next for Natural Language Processing 

It’s safe to say that the possibilities of NLP are endless, and it represents a huge market opportunity for anyone in the tech space. It’s also an area where startups and federal innovators may have a leg up on the tech giants like Amazon, Google, and IBM, who could lack focus. Industrious companies like U.Group are already forming niche understandings, uncovering challenges, and creating use cases that could become future proof of owning NLP verticals. 

Specifically, U.Group is exploring this by building NLP tools for the Beyond3 Marketplace. The ‘Marketplace’ is a collaboration ecosystem that uncovers funding opportunities and facilitates partnership between industry and the federal government. We use our NLP tools to validate data science models used in performing horizon scanning, which is the process of identifying where technology is maturing and who the industry/sector leaders are. In the most basic form, we do this by consuming mass amounts of data, cleaning the data, enriching the data, and then building algorithms to make sense of the data. With an NLG focus, we’re looking at how we can automate responses based on how humans respond, and improve over time, otherwise known as a stepping stone in machine learning…. which is a topic for another day.

Want to learn more about NLP or stay-in-the-know on U.Group products like Beyond3 Marketplace? Register for our monthly newsletter. 

Get alerted to new job postings, events, and insights by registering for our monthly newsletter.