Investigating the Efficacy of Ensemble Machine Learning Models in Multi-Class Categorization of Web Pages for Spotlighting User Interests
N Silpa, R Suneetha Rani, V. V. R. Maheswara Rao, N Amrutha, Shiva Shankar Reddy, Ramachandra Rao Kurada · 2024
Web page categorization plays a crucial role in organizing the ever-growing repository of online content, facilitating improved search functionality, enabling personalized content delivery, enhancing recommendation systems, optimizing targeted advertising, and spotlighting user interests. Utilizing the predictive capabilities of Ensemble Machine Learning (EML) models is essential for automating the web page classification process, ensuring a more efficient identification of user preferences in the dynamic landscape of online content. Towards this objective, the present research study introduces the Ensemble Machine Learning-based Webpage Categorization System (EML-WCS), concentrating on all essential stages of the classification process. The EML-WCS gathers a large and open-source dataset, ensuring a representative sample of web content across various key domains. The pre-processing phase entails a series of text processing steps, including replacing separators, removing URLs, punctuation, and numbers, eliminating stop words, and performing lemmatization to enhance the quality of the provided text data. The EML-WCS leverages the ensemble capabilities of Random Forest and XGBoost techniques for multi-class classification of webpages based on content, achieving an impressive accuracy of 99.69% and 96.04%, respectively. The outcomes of this study have implications for industries spanning from e-commerce to content recommendation systems, highlighting the pivotal role of understanding and responding to user interests for achieving optimal user satisfaction and engagement.