Natural language processing 101
As natural language processing transforms our ability to interact with machines, contributor Justin Freid takes a look at how it's being used and what marketers can expect to see in the future.
With buzzwords like AI and machine learning being thrown around more than a football at training camp, it’s important for marketers to understand the ins and outs of how the various components that make up these technology advances work.
Advancements in natural language processing (NLP) were one of the major steps forward that allowed AI to become mainstream. According to everyone’s favorite resource, Wikipedia, NLP is “a field of computer science, artificial intelligence and computational logistics concerned with the interactions between computers and human languages.”
In layman’s terms, NLP allows computers to understand what we are telling them.
As computers have become more powerful — increasing their ability to make mass amounts of computations analyzing inputs and learning how we “speak” to them — their ability to understand us has significantly improved. This is due to businesses seeing NLP, and ultimately, AI, as a business opportunity. Companies such as IBM have invested significant amounts of capital into programs like Watson that utilize NLP.
How and where NLP is being used
One of the areas that has utilized NLP for the last few years has been social listening tools. If you’ve ever run a social listening report and analyzed sentiment, this is a very basic example of NLP. The tools analyze the content within tweets, Facebook status updates and YouTube comments and determine if the shares were positive, negative or neutral. Basic NLP processing would highlight certain terms like “hate” or “sucks” as negative and terms like “awesome” and “love” as positive.
Many tools used a more advanced approach, being able to differentiate between the following:
“I got sick on the plane ride home.”
“I had so much fun — the roller coaster was sick!”
By looking at context and the other words used within the statement, social listening tools are able to differentiate between the two uses of “sick” and segment them into positive or negative shares. Many social listening tools have advanced this even further, segmenting shares to classes like anger, happiness, loyalty and other emotions.
As NLP has improved, so has the ability for social listening tools to understand what we share on social media.
Chatbots would not be here without NLP. By being able to comprehend your input, they can provide you with a quick answer that satisfies your request. Advancements in NLP have helped chatbots understand shorthand and misspellings to recognize the context behind your inputs.
This can be something as simple as a chatbot understanding that “ty” means “Thank You” and is not a misspelling of the word “tie.”
One of the most fascinating uses of NLP is Google’s RankBrain algorithm. As it constantly learns and evolves to provide better search results, NLP is utilized to understand the context behind your searches.
One example of this is how RankBrain has changed responding to queries such as “chinese restaurant near me.” Previously, you may have received results for Chinese restaurants in Maine due to Google flagging “me” as the abbreviation for Maine. Now, the algorithm understands “near me” is a location-based search and will utilize your physical location as part of the search.
One other area that has improved due to NLP has been interacting with voice-controlled devices. A perfect example of this is the new wave of technology in everyone’s homes, such as Amazon Echo and Google Home.
If you think back to how poorly Siri responded to voice commands when originally launched compared to how Siri performs today, the improvement is largely based on advancements in NLP.
Where will NLP take us?
As NLP continues to evolve and support AI, our ability to seamlessly interact with machines will only improve. No longer will we be frustrated because Siri does not text your friend the right message or a chatbot doesn’t understand your request to order a pizza with “extra cheez.”
Opinions expressed in this article are those of the guest author and not necessarily MarTech. Staff authors are listed here.