The HAR-KNN - A Novel Prospective for Recommendation System
Rohit Kumar, Myasar Mundher Adnan, Anurag Shrivastava, Shuchi Juyal Bhadula, Amit Dutt, K Praveena, Saloni Bansal · 2024
Commercial platforms frequently use the Recommendation System (RS) to offer recommendations to users. Recommendation systems (RS) are extensively employed in diverse domains, especially in e-commerce platforms where RS detects recommended products based on user activity. Some significant problems have emerged as a result of the last decades' significant growth in both users and products. Additionally, selecting the appropriate product and active user in RS is a challenging procedure. When recommending products, existing works take user preferences and sociodemographic behavior into account. Collaborative filtering (CF) is one of the most often used algorithms in recommendation systems. This algorithm is simple to understand and performs well. Data based on user behavior that is gathered from the Amazon dataset is used for testing. The Amazon product dataset contains 18,501 product reviews. The dataset's data was gathered and integrated with administrative services using Amazon Web Services. The suggested Deep Neural Network (DNN) method is empirically tested based on metrics like F1-measure, Recall (R), and Precision (P), and it provides the highest value of those metrics when compared to state-of-the-art techniques like K-Nearest Neighbour (K-NN) and Artificial Neural Networks (ANN).