A Hybrid Smartphone Recommendation Model Using Different Machine Learning Algorithms

Abstract

In today's world, mobile phones are an essential part of many people's daily lives. With several smartphones to choose from, people are often confused about which smartphone will be best for their use. This study attempts to present an accurate and usable prediction model for real-life mobile phone reviews of individual users to improve the model's prediction accuracy. A review-based prediction model is built on individual users' behavior and choice using machine learning. This data has been collected using web scraping tools like UiPath and the Python Beautiful Soup library from different websites, followed by data pre-processing. In this study, different Collaborative Filtering-based machine learning algorithms have been used and compared. The algorithms used include KNN based on individual items or individual users and an unsupervised SVD-based model. This has been demonstrated using UiPath Studio and the Druid AI chatbot. The Druid chatbot provides smartphone recommendations and data based on user input. Upon entering a smartphone name, a UiPath process is triggered, sending results back to the chatbot. This UiPath-based chatbot delivers specifications and recommendations. Future enhancements include broader product recommendations, improved user understanding through advanced NLP training, and an overall better user experience. Additionally, there will be a focus on incorporating user feedback to continuously refine and enhance the prediction model, ensuring it remains relevant and highly accurate.