Research On Recommendation Algorithm And Application Based On Particle Swarm Optimization

"625,000 videos were watched" and "51,000 apps have been downloaded on the App Store".These figures indicate that we are entering the era of massive data.While vastly growing data brings unlimited opportunities for business development,it also makes"information over-load" be increasingly fierce.Mining valuable information for users in massive information and data has become the major function of main-stream platforms.Recommendation system is an effective tool with this function,in which recommendation algorithm plays a key role.At present,recommendation systems and algorithms have been widely concerned and applied in many fields,such as e-commerce,video websites,personalized advertisements and so on.They have brought huge economic benefits for service providers.However,there are still many difficulties and challenges,such as data sparsity,accuracy and diversity,hindering the further development of recommendation.Particle Swarm Optimization(PSO)is a bio-heuristic optimization algorithm developed in the field of swarm intelligence computing in recent years.Because of its easy implementation,fast convergence and few parameters.PSO has shown its applicability and high efficiency in practical applications.Firstly,the basic principle and characteristics of PSO are analyzed,the shortcomings of it are summarized,and then an improved particle swarm optimization algorithm is proposed,named SA-PSO,which combined simulated annealing with basic PSO.Secondly,the idea and principle of semi-supervised clustering are summarized,and a semi-supervised clustering algorithm based on SA-PSO is proposed.Finally,the proposed clustering algorithm is applied to improve personalized recommendation,and an improved recommendation algorithm is proposed.On this basis of above work,a prototype of music personalized recommendation website is built.The feasibility of the improved algorithm is verifies to a certain extent.