Abstract :

Customer Helpdesks Handle A Large Volume Of Incoming Calls Every Day, And The Emotional Tone Of These Conversations Is One Of The Strongest Indicators Of Service Quality And Customer Satisfaction. Manually Monitoring Every Call For Tone And Sentiment Is Impractical, Time-consuming, And Highly Subjective, Which Often Causes Dissatisfied Customers To Go Unnoticed Until They Escalate Their Complaints Or Discontinue The Service. This Paper Presents An AI-Based Sentiment Analysis Of Incoming Calls On Helpdesk System That Converts Helpdesk Call Transcripts Into A Machine-readable Form And Classifies The Underlying Customer Sentiment As Positive, Neutral, Or Negative Using A Fine-tuned Bidirectional LSTM Deep Learning Model. Before Classification, The Call Text Undergoes NLP Preprocessing Such As Text Cleaning, Tokenization, Stop-word Removal, And Sequence Padding Using NLTK To Improve Prediction Accuracy. In Addition To Sentiment Classification, The System Estimates An Escalation Severity Level (Low, Medium, High) Based On The Intensity Of The Detected Negative Sentiment, Enabling Supervisors To Intervene In Real Time Before A Dissatisfied Customer Disconnects The Call. The Proposed Flask-based System With An SQLite Backend Enables Transcript Submission, Dataset Management, Model Training, And Sentiment Monitoring. The Bidirectional LSTM Model Achieves 87– 90% Prediction. The System Provides A Scalable, User-friendly, And Cost-effective Solution For Automated Helpdesk Call Quality Monitoring. Keywords: Sentiment Analysis, Deep Learning, Long Short-Term Memory (LSTM), Natural Language Processing, Helpdesk Analytics, Escalation Management, Flask, Customer Service.

Published:

01-8-2026

Issue:

Vol. 26 No. 8 (2026)

Page Nos:

1154 - 1162

Section:

Articles

License:

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

How to Cite