This article is the first of a series “hop on your Natural Language Processing Journey” about NLP:
- Chapter 1: Getting started with Natural Language Processing
- Chapter 2: email categorization
- Chapter 3: Create your first simple text classifier
- Chapter 4: Open-Source solutions vs provide
Joe works as an executive in the customer satisfaction department of a big company. As the person in charge, he often gets headhunted by companies claiming that they can help him improving customer satisfaction.
Usually, he doesn’t follow up those requests. This morning was different, a message aroused his curiosity:
“Embrace the natural evolution with Natural Language Processing”
“Natural Language Processing, what is that?”
Being aware of the technologies around him is part of Joe’s job. Of course, those words rang a bell to him.
- “So, NLP is the technology that helps computers understand human language, right?”
Touché. Natural Language Processing, shortened as NLP, is a branch of artificial intelligence whose objective is to read, understand and make sense of the human languages in a valuable way. Examples of NLP capabilities include analyzing sentences to structure them or extracting the subject of text documents.
“Chatbot? Not interested.”
But don’t try to fool Joe, he is used to deal with sales. With a hint of superiority that defines him, he was quick to articulate his counterargument:
- “Yes, I heard about it. We are talking about chatbots, AKA glorified decision trees that the clients always complain about.”
This is where you are wrong Joe.
The frustration around chatbots is understandable and while you are not wrong that the technology is not yet mature enough to have meaningful conversations with humans, NLP is much more than that.
When we think NLP, we often associate it with voice recognition and chatbots. Those technologies are useful and have substantially improved over the last few years. Think about voice recognition 10 years ago, and voice recognition through your car, or your connected devices now. But let’s not limit NLP to those cases with which we’ve all had some frustrations using it.
Right now, the technology comes in handy and is truly robust in many areas such as text classification, sentiment analysis, text extraction…
“What is in it for me?”
- “Okay I’ve got your point; NLP has made progress and might not be that bad in some areas. But how that can help me?”
Here comes the best part, you don’t need to be an expert to recognize the benefits of NLP and to find areas of high added value.
Here are some concrete use cases, but once again, you are invited to think outside of those boxes as the possibilities are endless:
- Improve Customer Satisfaction
NLP can be used to analyze customer messages or calls in order to determine the satisfaction of the customer using sentiment analysis. Insights about customer satisfaction can be used to get the sentiment about certain products or to help make target the right customers.
- Easy search
Finding the right information is key in lots of areas. Managing the indexation and retrieval of documents is made easier with NLP. NLP will help extract information on the documents as well as to make sense of the search query to make it understandable by the computer.
- Text classification
Classification of text involves automatically understanding, processing and categorizing unstructured text. It can be used for classifying the topics of documents or to extract the sentiments in text. Google gives us a concrete example by automatically using tags on website contents to get better search results.
- Text extraction
Text extraction automatically scans texts and extracts relevant information in unstructured data. Example of use case would be to extract information on contracts as a first step before indexing the documents in a database.
“Let’s give it a try”
Naturally, Joe being a rational person, he accepted to give it a try and after some discussions, he found a perfect case for NLP:
The customers’ requests are being handled manually by agents. They must read the messages and then format a reply. This process takes time and unnecessary effort. A way to improve it would be to implement an NLP solution that would automatically categorize the incoming requests of clients and propose a template of reply that can be modified by the agents.
Convincing Joe wasn’t that hard, a simple NLP analysis of his writing style showed someone curious and rational. Targeting the right people and adapting the tone of the arguments have proven to be highly effective in the process of finding new clients.
The easiest part is done, Joe is now convinced about the added value of NLP on this process, but are you?
Come next week to follow the implementation of email categorization and see the added value on a practical use case, that’s where the action begins.
Written by Charles-Antoine Vanbeers