Convolutional Block Attention Module based Wide DenseNet for Knowledge-based Recommendation System

Nowadays, the recommendation system is crucial to improve the product choices, which should satisfy the customer requirements in several domains such as e-commerce. To provide recommendations as per the needs of customers, deep learning based recommendation systems are utilized to enhance the accuracy significantly. However, due to changes in the customer needs the existing recommendation system based on neural networks provides irrelevant recommendations. To overcome these limitations, a Convolutional block attention module based Wide DenseNet (CW-DenseNet) is proposed for a knowledge based recommendation system to provide choices for customer needs. The proposed CW-DenseNet captures the features regarding the customer needs to provide correct product choices. The convolutional block attention module focused on the significant customer data to provide the accurate choice for the customers and enhanced the performance of the recommendation system. Experimental results of the proposed CW-DenseNet achieved accuracy of 0.886 which is higher than the existing recommendation system like Two Pathway Matrix Factorization (2 way-MF).