AM-LSTM: Academic Feedback Integrity Reinvented

The Vision Behind AM-LSTM

Student reviews are powerful tools — but also vulnerable to manipulation. Generic fraud detection models fail in academic contexts. AM-LSTM is developed specifically to protect educational feedback systems, ensuring transparent evaluations.

How It Works: The Science of AM-LSTM

Our model integrates an Attention-Modified LSTM for aspect-based sentiment extraction, followed by Random Forest classification for fraud detection. The hybrid approach ensures both deep contextual understanding and robust classification accuracy.

Proof of Excellence: Our Results

• Sentiment Analysis F1-Score: 86% (Own Data), 90% (Amazon Dataset)
• Fraud Detection Accuracy: 96.9%
• Precision: 99.9%
• F1-Score: 98%

Research Papers, Datasets & More

1. A. Bhowmik, Noorhuzaimi Mohd Noor, M. S. U. Miah, and D. Karmaker, “Aspect-based Sentiment Analysis Model for Evaluating Teachers’ Performance from Students’ Feedback”, AJSE, vol. 22, no. 3, pp. 287 - 294, Dec. 2023.
2. A. Bhowmik, Noorhuzaimi Mohd Noor, Md. Saef Ullah Miah, Md. Mazid-Ul-Haque, and Debajyoti Karmaker, “A comprehensive dataset for aspect-based sentiment analysis in evaluating teacher performance”,AJSE, vol. 22, no. 2, pp. 200 - 213, Aug. 2023.

1. A. Bhowmik, N. M. Noor, M. Mazid-Ul-Haque, M. S. U. Miah and D. Karmaker, ”Evaluating Teachers’ Performance through Aspect-Based Sentiment Analysis,” 2024 IEEE 9th International Conference for Convergence in Technology (I2CT), Pune, India, 2024, pp. 1-6, doi: 10.1109/I2CT61223.2024.10543706.