A Novel Multi-objective Evolutionary Algorithm Hybrid Simulated Annealing Concept for Recommendation Systems

Nowadays, recommendation systems have been widely used in information systems and internet applications. In order to solve the problem that most traditional recommendation algorithms mainly focus on accuracy and neglect other requirements such as diversity and novelty, multi-objective evolutionary algorithm has been introduced into recommendation system. But there is still much room for improvement in their performance. In this paper, we design a novel multi-objective evolutionary algorithm for recommendation system with accuracy, diversity, and novelty as its objective functions. In the proposed model, an efficient uniform distribution initialization and a mutation operator that integrates the concept in simulated annealing algorithm has been designed to improve the probability of producing more high-quality offspring. And an adaptive hybrid selection strategy is designed to select more valuable or promising individuals from offspring. Experimental results demonstrate the effectiveness of the proposed algorithm in terms of accuracy, diversity, novelty, and hypervolume compared to some state-of-the-art recommendation algorithms.