What you are about to read might frighten you, but the scare is worth the price of admission. To get into the Halloween spirit we decided to give you a rare look into how ReviewTrackers extracts spine-tingling business insights to improve the customer experience. Read on…if you dare.
A Crash Course on Natural Language Processing
The field of natural language processing (NLP) uses multiple techniques, including artificial intelligence, to power voice recognition, machine translation, chatbots, and much more. Think of it like Skynet from the Terminator movies except it doesn’t create killer robots (we think).
At ReviewTrackers, we use our proprietary NLP engine to extract important words and phrases from reviews and determine if the text in the review is written in a positive or negative light, a.k.a. sentiment analysis. This saves time compared to a human manually reading through hundreds of reviews to uncover trends in the customer experience.
Let’s say a customer left the following review:
It was hard to find without better signage. But it was worth it for their chicken carbonara.
Our NLP engine extracts the words “signage” and “hard to find” as negative sentiment and “worth” and “chicken carbonara” as positive sentiment. This type of analysis allows businesses to find ways to improve experiences for their customers, which can increase customer retention and drive more revenue.
Every day, our NLP models handle a wide variety of business types and review sites, but what if we used our technology on something more…unique? What if we used it on one of the most terrifying novels of all time?
For this Halloween season, we put our NLP engine to work on something quite new: the full text of Stephen King’s classic horror novel, The Shining.
Analyzing the Horror
Without any prior knowledge of the book, our model read the entire 400+ page novel and extracted eye-opening insights in less time than it takes you to drink a cup of coffee.
Warning: spoilers ahead! Read at your own risk!
We can determine the main characters of the book by seeing which keywords are extracted most often. This quickly reveals the names Jack, Wendy, and Danny. We then plot their sentiment over the course of the story.
In the beginning of the novel, the engine detected a mostly positive sentiment. The family is still happy together and Danny’s “shinings” have yet to play a pivotal role in the story. Here’s an example of keywords extracted from a few sentences early in the novel:
Jack smiled, a big wide PR smile.
Wendy is an extraordinary woman.
Danny was gleefully running his trucks across the sand pile
Once the family settles into the Overlook Hotel and the horrors begin, the book’s overall sentiment dramatically spirals into negativity. A huge contributor to that negative sentiment is the text surrounding Jack, who is now fully possessed by the hostile spirit of the Overlook Hotel.
“I said I’ll handle him!” Jack shouted suddenly, enraged.
Wendy was horribly pale
Danny was crying weakly.
However, sentiment doesn’t simply change whether a character is “seen” as good or bad; it also detects if they are in good or bad situations, which explains Wendy’s and Danny’s downward trend.
The NLP analysis shows an overall drop in sentiment from the beginning to the end of the novel. What starts as a tale about a seemingly normal family moving in to a hotel in the Colorado Rockies quickly descends into a story full of terror that could only come from the mind of Stephen King.
A Smart Engine for Your Reviews
The NLP engine’s analysis of The Shining is impressive, but imagine the impact it can have on your business and bottom line.
Identifying trends and specific keywords in reviews with the naked eye can be difficult. Layering our NLP engine on top of your reviews can provide valuable insights, which can turn into real-world innovations that significantly improve the customer experience.
Sentiment analysis provides immediate, rich insights to help tell a full story which saves an immense amount of time. Furthermore, it can help create context and uncover trends to improve experiences and increase overall business operations.
With ReviewTrackers, your business has an opportunity to seamlessly and efficiently uncover customer feedback trends, take action to build customer loyalty, and drive revenue.
More features are creeping their way into our NLP engine in the future. Our research team is working on a new algorithm to identify keyword trends – sudden changes in a keyword over time – which will allow users to identify trends or problems as early as possible.
You can see a preview of the new algorithm below where we added keywords selected by our trending algorithm onto the original chart. They are shown as “+” and “-” to indicate if sentiment is trending upwards or downwards. These words are associated with a specific section of the book, but not with any specific character.
The results get more frightening with each section, and align perfectly with some of the most iconic scenes from the story:
- Jack sitting at his typewriter trying to work on his book.
- Danny witnessing an elevator full of blood.
- Jack chasing his family with a mallet in the snowy topiary garden.
If you are interested in leveraging our proprietary NLP engine to gain rich insights into your customer experiences, click here.