Personalized Recommendation Algorithm Based on Data Mining and Multi-objective Immune Optimization

Abstract

In order to improve the accuracy of recommendation systems and user satisfaction, a parallel hybrid recommendation model combining data mining techniques and multi-objective immune optimization is proposed. A periodic intelligent recommendation algorithm framework combining data mining techniques and neural networks is proposed. This framework analyzes user historical behavior through collaborative recommendation and optimizes recommendation efficiency and speed using Top-N algorithm and grey balance algorithm. Meanwhile, combined with multi-objective immune optimization, this research method achieves a balance between improving recommendation accuracy and diversity, uncovering user preferences to provide diverse and novel recommendation lists. Combining data mining with multi-objective optimization, a hybrid recommendation system is constructed by prioritizing cascading and then parallelizing. The experiment showed that the proposed recommendation algorithm had a similarity accuracy of nearly 91%, which was at least 6.9% higher in accuracy than the two benchmark recommendation algorithms. The algorithm required fewer iteration cycles to achieve stable performance. In the testing of three different datasets, MovieLens, Donation Dashboard, NetfAix, the average accuracy of the personalized recommendation algorithm was about 95.2%, and the root mean square error remained below 0.04. Its performance was significantly better than other similar recommendation algorithms. The research method significantly improves the hit rate and normalized discounted cumulative gain value of recommendation results, providing users with more personalized resources that meet their needs.