Integrated Distance Based Convolutional Neural Network Optimized Using Elephas-Kiboko Algorithm for Movie Recommendation

A movie recommendation system aims to suggest movies based on users' preferences, past activities, and browsing history. These systems depend on algorithms that process extensive user behaviour data, movie attributes, and contextual information to provide personalized recommendations. This system suffers from several challenges like cold-start, sparse data, or dynamically changing user preferences. To overcome these issues, this research introduces the Elephas-Kiboko Optimization (EKO) Algorithm for the optimization of CNN based movie recommendation model. The proposed model also introduces an integrated distance metric that combines two standard distance metrics, namely Bhattacharya and Euclidean distances, to analyze movie similarities, considering both textual and visual features for enhancing the recommendation accuracy. The Elephas-Kiboko Optimization (EKO) method is a bio-inspired algorithm that utilizes clan-based updates and adaptive behaviors to enable effective navigation through the exploration and exploitation phases. This innovative approach addresses key challenges of recommendation systems and improves recommendation accuracy and user satisfaction. Experimental results on popular movie datasets show that the EK-CNN model delivers outstanding performance for metrics like accuracy, F1-score, precision, and recall with values of 94.87%, 94.65%, 94.00%, and 95.75%, respectively.