Analysis of e-Mail Spam Detection Using a Novel Machine Learning-Based Hybrid Bagging Technique.

Area of Science:

• Computer Science
• Artificial Intelligence
• Machine Learning

Background:

• Distinguishing between spam and non-spam emails remains a significant challenge for both providers and consumers.
• Existing spam identification techniques require further enhancement in accuracy, training time, and error reduction.

Purpose of the Study:

• To propose a novel machine learning-based hybrid bagging method for improved email spam identification.
• To develop a framework that accurately categorizes emails as either spam or ham (non-spam).

Main Methods:

• A hybrid bagging approach combining Random Forest and J48 (decision tree) algorithms was developed.
• Preprocessing involved tokenization, stemming, and stop word removal, followed by Correlation Feature Selection (CFS).
• The model's effectiveness was evaluated using metrics such as accuracy, precision, recall, and F-measure.

Main Results:

• The proposed hybrid bagged model achieved a high accuracy rate of 98% in email spam identification.
• The method demonstrated superior performance compared to existing techniques in reducing error rates and training time.

Conclusions:

• The novel hybrid bagging method offers a robust and accurate solution for email spam detection.
• This machine learning-based approach significantly enhances the ability to differentiate between spam and legitimate emails.