Application Research of Intelligent Algorithm in Short Video Recommendation System Optimization Personalized Recommendation System

This project aims at high relevance, low interference, short and precise, and researches the deep network target recognition and recommendation method for short video targets. In view of the differences in the source and network environment of short video, two realization methods of cloud computing are proposed: cloud platform and mobile terminal platform. The model consists of a server, a content delivery network (CDN), and a terminal. The server can discover and confirm the target of the CDN in advance. It matches the corresponding advertising content to achieve mobile terminal broadcast. Through the analysis of local video, the target detection and recommendation are realized. In order to overcome the shortcoming that the existing LFM method cannot recommend users immediately, a short video recommendation method based on LFM is proposed. When the user gives positive feedback on a particular video, the ItemCF method is used to search for N videos that are closest to it but have not been watched. Then LFM algorithm is used to generate the corresponding recommendation table, which is integrated with Redis cache library to achieve instant short video recommendation. The combination of Java and C++ is used to improve the speed of computation, and the feedback mechanism is used to reduce the number of target classes to improve the real-time computation.