3 December 2020

Trending Use Cases Where AI Fabric OOTB Models Can Fit In

3 December 2020

Trending Use Cases Where AI Fabric OOTB Models Can Fit In



Editor's note: As the automation market continues to evolve, the UiPath Platform also updates to best serve the automation needs of our customers. As such, some of the product names in this article have evolved since the article was originally published. For up-to-date information, please visit our AI Center page.


Back in January 2019, as a robotic process automation (RPA) developer with Zensar Technologies, I was working on a project which required a text classification machine learning (ML) model, and the integration of Python and RPA. It was really challenging to learn Python and implement it as an executable method with RPA.


When UiPath launched UiPath AI Fabric, I was really mesmerized with its features. I didn't even have to use Python or data science for creating such text classification models.


RPA and artificial intelligence (AI) both are trending technologies in the market. Think about how many human hours can be saved when these two technologies come together and work on a single platform for making your life easy. UiPath has really proven “What used to be a science fiction is now a science fact!”.


The interesting part of AI Fabric is that you actually have more than 20 pre-defined, out-of-the-box (OOTB) packages. Which means you don’t have to write a ML project, train an ML model, and work on data pre-processing because everything is being taken care of.


Today in this blog we are going to look at:

  • How can we leverage an OOTB package to grow your business with AI and RPA?
  • What are the use cases we can automate with the help of UiPath AI Fabric?



Before we start with the models, find out what AI Fabric is first. Then, watch AI Fabric bridging the gap between RPA and AI: 



Sentiment Analysis ML model

Description: This model predicts the sentiment of a text in English Language. Possible predictions are one of "Very Negative," "Negative," "Neutral," "Positive," "Very Positive."

Input to this model is a simple English sentence: “I am dissatisfied with the service.”

Output of this model is a sentiment predicted: {"sentiment":"Very Negative","confidence":0.97}


Let us look at what are the possible use cases where Sentiment Analysis will fit in.


1. Identifying customer feedback sentiments for an e-commerce business

This is one of the most common and most practiced use cases for Sentiment Analysis where we can identify sentiments from customer reviews on a product review platform. The model can be used to find patterns and identify client sentiments for product reviews on major e-commerce websites such as Amazon, Flipkart, Alibaba, eBay, and Rakuten.


2. Finding patterns of seasonal and regional impact on product reviews

We can identify patterns of seasonal impact on product reviews. For example, we can identify sentiments from customer feedback on a clothing collection in winter. And we can compare it to the feedback on clothing collection in summer. It will help clothing shops to understand the seasonal impact of clothing collections on customers.


Also, we can find patterns of regional impact on product reviews. For example, we can generate data from a survey to verify the need for a pizza store in an urban area by quickly identifying sentiments from the survey data. And the same survey and analysis can be done for a rural area.


We can leverage these patterns identified from sentiment analysis to improve our business. It helps us market our products to exact regions and saves a huge amount of time spent on reviews and survey analysis.


3. Identifying target audience

Many companies spend huge amounts of money on surveys and campaigns to understand their potential customers (who have visited their company websites or social media accounts). Once the survey is done, it again takes a huge amount of time to identify rightful customers and find out hidden patterns in the feedback.


With the UiPath AI Fabric OOTB Sentiment Analysis model, we can identify the exact target audience (in a certain age group, for example) based on customer reviews and categories of customers. People in some age groups tend to have positive feedback on a certain product.


4. Employee feedback analysis

Every company is built on top of its employees. If each employee has job satisfaction, then the company can excel in all aspects. Hence understanding employee’s feedback and improving based on that is really a critical and important factor for each organization.


Companies usually do a yearly “Great Place to Work” survey. It takes a huge time for an organization to go through each survey feedback and analyze it. The Sentiment Analysis model will make it faster and easier to identify a class of “Very Negative” feedback and address those employees quickly within any huge feedback data set. It saves a lot of time and helps organization to maintain a healthy environment and culture for employees.



5. Identification of fraudulent brand influencers

Sometimes product reviews are intentionally manipulated by fraudulent influencers. However, it becomes very difficult for organizations to identify this scam and fraud.


Sentiment analysis can help in identifying negative review patterns. It can also help organizations to focus on only a target set of reviews and feedback on social media where brand value is getting impacted. Once the pattern or repetition is identified in this data, we can detect fraud easily.


6. Competitive analysis

We can leverage Sentiment Analysis in identifying competition research analysis by identifying sentiments from market research and keeping a closer eye on a competitor’s product feedback. We can utilize this data to improve or innovate based on market requirements and excel in competition.


For example, Company A and Company B both sell laptops. Company A keeps a closer eye on Company B’s feedback portal to find insights on what can be improved and what can be avoided based on customer feedback sentiment. This helps Company A maintain healthy competition.


7. Reputation monitoring

Sentiment Analysis can be used in identifying sentiments from social media engagement to monitor reputation of a product, a company, a service, or a brand. Identifying sentiments and keeping a closer eye on negative sentiments helps in monitoring what could go wrong with a latest decision.


People have habit of sharing their emotions in comments, feedback or any social media posts. We can leverage such data related to hashtags or a brand name to identify and monitor the market reaction to the decision, and the actual reputation of the brand from customer’s perspective.


8. Client retention management

The biggest advantage of the Sentiment Analysis model is improvisation in customer centricity. For every business, customers are the center position of all, and the entire business is always around what customer needs and how to maintain customers. The biggest challenge to every product is customer retention and knowing customers’ needs.


With Sentiment Analysis, one can identify sentiments of valuable customer feedback and make sure to provide better service to those who have negative feedback.


Retaining customers is one of the biggest challenges which can be improved with knowing your customer better. For example, a shopping complex automates their customer feedback documents to identify sentiments from customers about shop visits. Based on results, the shopping complex will send out an apology email with vouchers or coupons for next visits and send thanks emails to shoppers with positive sentiments.


Nisarg Kadam is an RPA Center of Excellence (CoE) Lead at CRG Solutions.

by Nisarg Kadam

TOPICS: Artificial Intelligence, RPA Use Cases, RPA + AI, AI Fabric

Show sidebar