Sentiment Analysis: How can it make AI smarter?

Written By : Prof. Hobby
Date Posted : Feb 27, 2019

“You are a real boy. At least as real as I’ve ever made one.”

– Prof. Hobby to his AI robot, David (Characters from 2002 Hollywood movie A.I.: Artificial Intelligence)

An independent survey suggested that 56% of the CIOs of Insurance companies quote that multiple business issues across the value chain can be solved by “improving operational efficiencies”. An independent survey suggested that 56% of the CIOs of Insurance companies quote that multiple business issues across the value chain can be solved by “improving operational efficiencies”.

CONTEXTUAL UNDERSTANDING

By observing the behavioural pattern of humans, we can make sense of the person’s mental and emotional condition at any given time. This will then influence our response, appropriate to the best of our knowledge. Consider the following sentence:

“My bank rejected my loan request. Amazing!”

Clearly, to any human, the above statement is in a bad taste. The word ‘amazing’ makes the statement sarcastic, which is again a negative emotion. Most humans can apply contextual understanding to this statement and comprehend the true sentiment behind the expression. For a machine though, it will catch the word ‘amazing’ and term the statement as a positive one.

This is the difference between humans and machines.

Imagine the customer’s frustration if he gets an automated reply saying, “We are delighted to know about your amazing experience with our services, and we look forward to serving you again!”

Allstate launched chatbots which enable agents to learn about different ways to sell commercial insurance products for the first time. It gives the agent a walkthrough of the entire process, helps them if they are stuck in the middle and can even help in retrieving documents.

Indian insurance player HDFC adopted AI tech to run a questionnaire to help people solve their queries related to insurance via mobile chatbots platform. This enabled them to handle a huge amount of queries with minimum errors, and quick response time.

Another example of how AI can be used in Insurance space is to transform user experience using a speech recognition application combined with artificial intelligence that will enable customers to check balance premiums, claim status and solve all insurance-related queries at ease.

SENTIMENT ANALYSIS WITH AI

The past year has witnessed a lot of hype and exponential growth in the implementation of artificial intelligence in various sectors. Machine learning algorithms are created to automate various processes within the Sales, Marketing and HR functions. The implementation of AI technology has certainly helped organizations in improving the top line.

Deep learning has improved the machine’s ability to understand a query and give an appropriate response. The companies can find the opinions and views of their customers on various social media channels such as Twitter, Facebook, YouTube, etc. This data is fed through machine learning algorithms to understand the customer sentiment (positive, negative or neutral) to provide an appropriate response. This happens so smoothly in the background that a customer posting the query might not know that he is speaking to a Bot unless explicitly mentioned.

As mentioned in the example above, it is evident that the customer’s feedback on getting his loan rejected is a negative one and the situation could go down south if it is not handled properly.

Apart from the quick response functionality and the ability to give accurate resolutions, the companies have shifted their focus on AI tech to make the automation smarter. The most basic use case would be an intelligent auto-response to an irate customer posting negative statements on chat or social media. While most of these comments can be handled correctly by the automated systems using an IF-THEN algorithm, it is the sarcastic comments that could be tricky. So, the system needs to ‘understand’ that if a loan is rejected, it is not a pleasant experience for the customer regardless of the adjectives that follow. The machine learning techniques and natural language processing have an important role in identifying sentiments from the customer’s contextual statements. Using text mining, i.e. extracting high-quality information from the text and through text analysis, systems can be trained to handle mocking or sardonic queries with the right approach.

Beyond everything else, realistically the human sentiments cannot be divided into just three buckets of Positive, Negative and Neutral. With the advancement of technology and social platforms, language is evolving faster than ever before, and the AI must keep up with the pace. Exponentia.ai is an AI tech provider that understands the importance of building an intelligent system that not only makes automation smarter but also acknowledges the human expressions.

MACHINE LEARNING

Technology

LEAVE A REPLY

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

POST COMMENT