What is Natural Language Processing?

NLP is a subset of AI which allows computers to process (in this case process means understand) words in the same way humans can. NLP includes both text and spoken language.

​

​

Computer Language

 

Unlike humans (and other living beings), computers understand a language called binary code. This is represented as a series of 1s and 0s (Have you ever heard people on TV says "it's all 1s and 0s" or something similar? They are talking about binary).

​

taught - 01110100 01100001 01110101 01100111 01101000 01110100 

by - 01100010 01111001 

humans - 01101000 01110101 01101101 01100001 01101110 01110011 

​

Obviously, human brains do not typically (or quickly) process or think in binary code. Therefore, some very clever people inventing coding languages (such as Python, C, Java) to be able to communicate more effectively with computers. Python code to print Taught by Humans:

​

print("Taught by Humans")

​

The aim of NLP is to allow humans to communicate with computers without needing to write code, and for it to feel more like the natural language which they use to speak to other humans.

 

Use Cases for NLP

​

• Chatbots

• Sentiment analysis

• Spam detection

• Translation

• Summarising text

​

Interesting Examples

​

Virtual Assistant - Alexa and Google Assistants

NLP has allowed virtual assistants to become more human-like, and therefore more useful. If anyone has been using an Alexa for years, the advances in NLP has created a better experience in many ways over the years, for example understanding more accents, less misunderstood commands.

​

ChatGPT

The most recent (and potentially famous) example of NLP has been ChatGPT. As of August 2023, ChatGPT 4 is the most advanced example of NLP. It understands and responses like a human (even if it sometimes feels a tad off or robotic - we were going to add incorrect, but that is a very human trait).

​

Google Translate

Google translate is a commonly used tool powered by NLP. The advances in NLP can be shown as Google translate used to only directly translate languages (using a method different than NLP, such as looking up a dictionary with no understanding or context). Some examples of developments in Google translate using NLP:

• Translating Spanish to English, "tomar" is translated as "take", however "tomar y comer" is "drink and eat". Using human like understand to determine the word has different meanings in different situations.

• Another example, taken from the IBM website, using English to Russian and back again. Previously this changed “The spirit is willing but the flesh is weak” to “The vodka is good but the meat is rotten.” Now this seems to return the correct phrase when translated to Russian and back once, but if translated more than this "is willing" is changed to "wants" so quite close.

​

Related Resources

What is AI?