Workforce Experience Indicators derived from Employee Text under Transformer-Based Multi-Label Analysis

Abstract

Real-time statistics of Amazon employees reveal several patterns in workload, job satisfaction, and workplace challenges. These insights help understand the day-to-day experiences of employees within the organization. The motivation behind studying these aspects is to identify the major factors that influence employee well-being and overall performance. It also highlights the need for a systematic approach to understand workforce-related concerns. This research focuses on collecting, reviewing, and analysing Amazon employee data to identify common issues faced by the workforce. The study aims to present clear indicators of employee experience and provide a structured way to interpret and understand the problems observed among Amazon employees. This study uses (sentence-Bert) SBERT-based text representations combined with the Stochastic Gradient Descent (SGD) method to analyse employee reviews effectively. The SBERT–SGD combination supports multi-label classification and helps extract relevant workforce experience indicators from Amazon employee’s data. This study focuses on systematically collecting, reviewing, and analyzing employee reviews to identify common workforce-related concerns. The goal is to highlight clear indicators of employee experience and present them in a structured and understandable manner. Text-based representations are used to examine employee feedback effectively and interpret recurring issues observed among Amazon employees. The approach supports multi-category classification of workforce experience indicators and helps extract meaningful insights from large volumes of employee data.