A Machine Learning-Based Double Token Weighted Clustering Approach for Online Product Recommendation

Abstract

Machine learning techniques have generated significant interest in the development of recommender systems. Given the vast array of direct and indirect variables that can be used to predict user preferences, there is a growing need for scalable, reliable algorithms and systems that offer high availability and scalability. In today’s technologically advanced era, people have become more open-minded and increasingly depend on modern applications for daily needs such as purchasing accessories, watching movies, and more. The rising demand for online shopping and media consumption has led businesses to adopt machine learning-driven technologies to efficiently identify the most relevant products for users, with less effort compared to traditional marketing methods. Recommender systems (RS), particularly content-based filtering systems, play a vital role in both personal and professional contexts. These systems act as intermediaries between content providers—including social media platforms, e-commerce websites, and streaming services—and end users by suggesting items that match user preferences and past behaviors. Such personalized solutions are especially valuable when users are uncertain about what they want. Clustering is a key method in this space, which involves organizing a population or dataset into distinct groups, ensuring that data points within a single group are more similar to each other than to those in other groups. The goal is to identify users with similar characteristics and group them together in clusters associated with specific products. This research introduces a Double Token Weighted Clustering Model (DTWCM) designed to analyze and group relevant product recommendations sourced from multiple online recommendation systems. The model efficiently delivers high-quality suggestions to users. When compared with the traditional Adaptive Weights Clustering model, the proposed DTWCM demonstrates improved accuracy and scalability.