The Social World today has ‘sentiments’ waiting to be tapped. Take the case of this brand that used social listening and sentiment analysis to make better decisions in improving the sales of products. Some customers pointed at the ‘weak’ features of products, some highlighted the plus of ‘on-time’ deliveries and some even expressed opinion about the pricing of products.
So much for the brand to gain intelligence from sentiment analysis. And as the brand wanted to take action based on the intelligence, it lacked the recommendation system to take desired actions and achieve successful business outcomes. The brand finally found the cohesive unit in Sentiment Analysis and Recommendation System.
How to move from sentiment analysis to recommendation system?
Sentiment analysis and possibilities
Sentiment analysis also recognized as Opinion mining is a subfield of Natural Language Processing (NLP). For analysing text data, especially with the need to understand opinion, human interaction is leveraged at all times. But with the advent of NLP, machines are now able to understand human language and there has been a surge in the usage of textual data for Text Analytics.
Before deep diving into the complete picture of the Sentiment Analysis, let us look at some wonderful use cases
- Have you ever raised an issue at one of the leading e-tailer’s Customer Care point, and in case you have, then you will be stacked in a priority list. In unearthing the How, usage of Sentiment analysis in understanding the highest priority of customers becomes clear as a crystal.
- Have you ever been stuck at the Airport because of the flight running late?
And do you ever register a complaint?
But for all this, excuses are the only answers you get from the airline people. Now, accessing the Twitter account of the airlines company and just dropping a tweet expressing your anger will evoke a speedy response.
Now for the ‘Why’, it is Thanks all the way to Sentiment Analysis that they are leveraging. It calculates your opinion and gives out a value.
- Have you ever wondered that a person with suicidal thoughts or a person affected by depression can be identified and prevented from committing such irrational acts just by running sentiment analysis on their social feed?
- And last but not the least, brands that have Twitter handles are keenly mining customer opinions on Twitter and modelling their products based on the intelligence from sentiment analysis.
And there are many use cases that go on from Scalability, Consistency to Real-time Interactions.
Moving to Recommendation System
The real big question that arises is who are the customers that need to be focused?
It augurs well as not to fall into the trap of customer reviews and model brand decisions and actions based on that. Looking at just the ‘customer stars’ turns inadequate in mining intelligence, triggering the need to dig deep to facilitate better company modelling and evaluation.
Saksoft’s engagement with a client gave the opportunity to make a deep analysis of data, and most importantly find a way to detect customers who needed more focus rather than just resting faith on customers’ stars.
As dictated by human behavior, people like to respond in the form of text rather than just give out start ratings – when customers really like something, and when they don’t approve something, customers express it in terms of action, with Text being the medium. By combining text and star reviews, brands can potentially sharpen their focus on customers that deserve focus.
Further, while using dataset of fields having the recommendation score, positive comments and feedback comments, all available data can be categorized into two segments:
- Good feedback
- Improvement feedback
Applying text pre-processing and then applying sentiment analysis on text yielded scores ranging from -1 to 1(setting our own threshold, assuming 1 to -1). Now, the one with the highest score meeting the threshold one are the customers who have given honest feedback with good measures. Thus, a brand can find their strengths and can focus more on their strengths.
Going through the same process of preprocessing and then applying sentiment analysis, focus turns towards the bottom score meeting the threshold of -1. This alludes to customers offering comments, who are the ones that are not impressed by the company and pinpoint shortcomings of the brand. By taking a closer look at these mentioned comments, brands can find their short comings and find a proper solution to meet customer needs and improve business.
And when it comes to Recommendation system, customer opinions, thoughts and expressions come up as the prime feed. Leveraging customer feedback, brands can come up with the robust Recommendation system to optimize services, products and the business on the whole.