Sentiment Analysis Solution
For employee Engagement Platform

INDUSTRY

HR Tech

CORE TECHNOLOGY

Python

About Client

A Canadian company that provides SaaS based employee recognition solutions to boost employee engagement in modern enterprises.

The company is working in this domain for more than 100 years, having offered employee recognition solutions to global organizations with a workforce between 10000 to 85000 employees.

The solution provider was looking to revolutionize their employee recognition process with advance intelligence. With the same vision in mind, the client collaborated with Azilen to develop an intelligent sentiment analysis solution for its proprietary employee engagement platform.

Challenges

The client was looking for advance sentiment analysis solution for its existing SaaS based platform. Looking to infuse an intelligent sentiment analysis module in an existing platform, the client wanted to:

  • Develop custom solution as there are many players in the market but not offering advance insights
  • Gauge & evaluate employee sentiments from 50,000+ emails & lakhs of real-time entries/records on their platform
  • Evaluate each comments/emails/records to identify and categorize opinions expressed in a piece of text
  • Base on sentiment of each opinion; solution decide rating itself

Solution

Azilen understood the client requirement in-depth and adopted supervised learning approach to provide an agile and sophisticated sentiment analysis solution. Robust machine learning libraries were identified to understand how humans interact and categorize the input into positive or negative sentiments.


To develop a powerful solution, it was important to understand the employee recognition scenario, devise a plan to transmit the crude information into valuable data and deliver the final results after sentiment analysis. The solution passed through the following stages:


    Scenario Analysis

  • Managers wrote recognition messages for team members periodically to laud their achievements and performance.
  • Employees provide feedback on manager role through regular surveys
  • The information is recorded in English language and is available through API

    Development of Data Integration Service (DIS)

  • Transmits crude data (information) accessed via an inbound web API
  • Extraction, cleaning and organization of crude data
  • Transmission of organized data to prompt ‘sentiment analyzer’

After the above process was identified, a sentiment analyzer was developed with the following capabilities:

    Sentiment Analyzer (SA)

  • Activated on trigger generated via DIS to prompt sentiment analyzer
  • Process the available data using NLP to identify, analyze and categorize positive or negative sentiments
  • Analyze employee survey ratings
  • Establish co-relation between the NLP-based sentiment analysis and survey ratings to come up with consolidated rankings
  • Corroborating dataset rankings with positive/negative sentiment percentage
  • Transmitting the final rankings back to the DIS

Tools & Technologies

  • - (NLTK)
  • - Gensim
  • - Scikit-learn
  • - SpaCy
  • - (NLTK)
  • - Gensim
  • - Scikit-learn
  • - SpaCy
Benefits of Automated Response Capability Integration
  • 20% improvement in employee recognition process
  • 90% accurate sentiment analysis
  • 100% increase in productivity