A MACHINE LEARNING APPROACH FOR DETECTING OVER-APOLOGIZING IN PROFESSIONAL COMMUNICATION

Abstract

Apologies play a crucial role in maintaining politeness and professionalism in workplace and customer support communication. However, frequent or unnecessary apologies referred to as over apologizing may convey insecurity, reduce perceived confidence, and weaken the tone of communication. This research presents a machine learning-based framework for automatically detecting apologetic expressions and identifying potential over-apologizing behavior in workplace emails and support responses. Using the Enron Email Dataset, extensive text preprocessing and Term Frequency–Inverse Document Frequency (TF- IDF) vectorization were applied to extract meaningful linguistic features. Multiple models, including Logistic Regression, Naive Bayes, and Random Forest, were trained and evaluated for apology classification. Logistic Regression achieved the highest accuracy of 90.3%, demonstrating its effectiveness in detecting apology-related patterns. The study highlights how Natural Language Processing (NLP) techniques can be leveraged to analyze communication tone, promote assertive writing, and provide insights for professional and organizational communication improvement.