How NLP is Bringing Humans and Computers Closer

A look at how natural language processing (NLP) can improve the way we interact with computers.

In today’s world, where any information we want is a click away on Google, it is frustrating that many professionals still have to manually work through text documentation to do their jobs. Lawyers, for example, can only build out cases after working through piles of evidence to find the information they need. This is inefficient and slows innovation down. 

With artificial intelligence becoming more commonplace, however, we are starting to see solutions to this problem being developed. Natural Language Processing (NLP) is quickly emerging as a powerful tool that can enable computers to process and analyse large amounts of natural language data, making human and computer interaction easier. This is a game changer for professionals, like lawyers, because it removes a significant amount of manual effort on their side, which allows them to focus their attention on cracking cases, not reading through documentation. 

Here’s how NLP works in more detail and why it can be such a powerful tool.

What is NLP and how does it work?

Natural Language Processing is a tool that essentially allows us to engage with our computer as we would with another human being. Because it has the ability to process and analyse a large amount of ‘natural language data’ – or information that has been captured in the same way that humans think and speak – it can understand and respond to our requests in a natural way. 

> This is powerful because it allows us to communicate with our machines in a direct way, which means quicker feedback and results.

In many ways, we are familiar with being able to ask and receive from our computers: Keyword search and online search engines are tools that we interact with every day. However, to use these tools well, we need to be specific about what we ask from them. Keyword searches will generally only show exactly what is searched for meaning results will be limited, and while search engines like Google will show more expansive results, they aren’t exact and still require us to work through content to find the details we want. This limited capability means that humans are forced to communicate with the computer in more iterative ways, continuously refining what we ask to get the results that we want. We have to plug in words in a certain order, for example, that doesn’t mirror how we speak.

> If computers exist to help make human’s lives easier, there shouldn’t be a ‘language barrier’ that slows interactions down. 

NLP, on the other hand, makes this process much more intuitive, which means that there is more freedom and flexibility with how we can interact with computers. 

Instead of doing a somewhat limited search on specific keywords that will likely bring back tens or even hundreds of results, a lawyer – for example – could use NLP to conduct a ‘semantic search’ using phrases. This returns a ranked list of sentences that match the search phrase – even if none of the specific words used in the search phrase are present in any of the documents. This might sound like magic, but because the technology has been trained to analyse information as humans can, it ‘understands’ the words in the search phrase and how they relate to each other – and then uses this understanding to interpret what we are really asking for and return a more nuanced result.

For example, there are many words in the English language that sound or even look the same but have completely different meanings. The word ‘date’ is one: It could refer to the fruit or to a specific day. If our lawyer were to search for, “what date did the second hearing take place,” using semantic search, the computer would be able to recognise that they were referring to the day, the same way a human would. It would then proceed to extract a similar ‘understanding’ and further derive meaning from other words in every sentence it searches through to find the closest match.

NPL in the real world

Research is one field where NPL technology could have a profound impact. When COVID-19 struck, Kaggle released a data set that contains 44 000 research papers conducted on the virus. Our team at Deep Learning Café saw the opportunity to work with NLP to build an explorer that could use semantic search to return the five most relevant papers from the data set. With so many changing theories and fake news surrounding the coronavirus, having a tool available that can constantly keep up and return what is relevant is extremely useful for staying informed. You can see the results of our explorer in this video

In closing, semantic search tools can be extremely powerful for anyone looking to become more effective at what they do. By adding depth to how our computers ‘understand’ text, and therefore changing the dynamics of how we interact with them, we are working towards saving time, cutting costs and performing more accurate work.


Tawanda Ewing is an Electrical & Electronic Engineer turned Machine Learning Engineer with a passion for spreading knowledge and a desire to become an Artificial Intelligence expert. Developing the skills to solve real world problems, particularly in Africa, is his main motivator and he believes that a solution to every problem is always within reach when individuals with the same vision form a team.

Leave a Comment

Your email address will not be published. Required fields are marked *