What is Natural Language Processing? A Beginner’s Guide
For computers, our natural form of communicating is incredibly hard to grasp: enter Natural Language Processing.
What is Natural Language Processing?
Natural Language Processing (NLP) is a subfield of AI that helps computers understand natural human language. Its main goal is to improve human-machine communication. AI's goal is to create machines that can perform some of the cognitive tasks that humans can. Among these is the ability to understand each other through languages that have evolved naturally, without premeditated planning. When we interact with each other, we use a range of verbal and non-verbal signifiers that constantly evolve. For computers, which are programmed with constructed and formal languages, this natural form of communication is incredibly hard to grasp.
How NLP makes sense of what we say
This is where Natural Language Processing (NLP) comes into play. NLP is a subfield of AI, linguistics and computer science concerned with improving understanding between man and machine. It is also a sub-form of machine learning, which uses algorithms and statistical models to enable computers to perform tasks they were not programmed for. Applied to NLP, machine learning lets machines absorb vast amounts of natural language data and make sense of it in two main ways: syntactic analysis, where syntax and grammatical rules are used to derive meaning from a text, and semantic analysis, where algorithms focus on what the words themselves mean and how they can be interpreted depending on context.
NLP is everywhere
NLP is already used in many everyday applications like autocorrect, predictive typing, chatbots, translation apps, and personal assistants like Siri or Alexa. NLP is also used by HR professionals to sort through applications and resumes, by BtoC companies to sum up the salient points in written customer feedback, or by publishing companies for more accurate news and content aggregation. Overall, AI understands us very well: a tool developed by Google in October 2018, called BERT, outperforms humans on advanced reading comprehension tests. But it's unclear whether the machines are getting better at truly understanding us or at playing the game.
"We know we're somewhere in the grey area between solving language in a very boring, narrow sense, and solving AI."
Sam Bowman, a computational linguist at New York University, told Wired. But even the "boring, narrow" option is full of potential: according to Gartner, by 2020, 85% of customer interactions will be managed without humans. Talking to machines will be the most natural thing in the world.
Stay in the loop
Subscribe to our email newsletter. No spam, just occasional insight from our experts.