• Umar Farooque
• Dr. Puneet Nema
• Dr. Vivek Richhariya

Spam detection is an important natural language processing (NLP) task that aims to accurately identify and extract relevant text spams from unstructured textual data. However, conventional machine learning approaches often face challenges in capturing contextual relationships and semantic dependencies within complex language patterns. This study proposes an optimized spam detection framework integrating Bidirectional Encoder Representations from Transformers (BERT) with the XGBoost technique. BERT is employed to generate context-aware textual representations, while XGBoost is utilized to improve classification and prediction performance through efficient gradient-boosting learning. The proposed BERT-XGBoost model was evaluated using standard performance metrics, achieving an accuracy of 96.93%, precision of 96.14%, recall of 90.88%, and F1-score of 93.43%. Furthermore, the model obtained an excellent ROC-AUC score of 99.50%, demonstrating its strong capability to distinguish relevant and irrelevant spams. The obtained results indicate that the combination of deep contextual representations from BERT and the robust classification capability of XGBoost provides an effective and accurate approach for spam detection. The proposed framework can be applied to various NLP applications requiring reliable extraction and classification of meaningful textual spams.

BERT, XG Boost, Accuracy, Precision, F1-score

"Optimization Accuracy of Spam Detection using BERT and XG Boost Technique", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.13, Issue 9, page no.a401-a407, September-2026, Available :http://www.jetir.org/papers/JETIR2609047.pdf

"Optimization Accuracy of Spam Detection using BERT and XG Boost Technique", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.13, Issue 9, page no. ppa401-a407, September-2026, Available at : http://www.jetir.org/papers/JETIR2609047.pdf