Collaborative Filtering Recommendation Algorithm Based On Improved NSGA-Ⅱ

While the rapid development of the Internet has brought convenience to people,it has also brought many problems.Faced with a large amount of information,people often do not know how to choose.The recommendation system is a tool that helps users quickly find useful information.It is a personalized system that is "tailored" for users.It can make item recommendation based on the user’s preference demand model.In this process,items that more closely match the user’s preference needs are more likely to be recommended to the user.Collaborative filtering algorithm is one of the most classic and successful recommendation algorithms.The similarity measurement method of traditional collaborative filtering recommendation algorithm ignores the problem of behavior consistency among users,resulting in inaccurate similarity calculation.In addition,traditional recommendation algorithms pay too much attention to the accuracy of the recommendation results,and ignore the importance of mining long-tail items.In the process of recommendation,whether it can improve the diversity and novelty of recommended items without reducing the accuracy rate has become the focus of attention.But the recommended accuracy and diversity are two contradictory performance indicators,which requires a trade-off between the two.Multi-objective optimization algorithm is a method to find the optimal solution on multiple targets.Genetic algorithm is a search algorithm that simulates the biological evolution process to solve multi-objective problems.The genetic algorithm is applied to the recommendation system.This is accuracy-diversity.Dilemmas provide a new way of thinking.Aiming at the above problems,this paper first designs a similarity calculation method based on user behavior consistency.This method rewards or punishes similarity calculations based on whether user ratings are consistent,and considers the degree of dispersion of user ratings and the intersection ratio of user rating items in the similarity calculation.Then,a traditional collaborative recommendation filtering algorithm based on improved NSGA-Ⅱ is proposed to solve the problem of only focusing on the accuracy of recommendation results and ignoring recommendation diversity in traditional recommendation techniques.The traditional recommendation problem is modeled as a multi-objective optimization problem.A method for dynamically calculating the crowded distance is proposed for the shortcomings in the NSGA-Ⅱ algorithm,and a cross mutation method based on the close relative coefficient is designed.Finally,the improved NSGA-Ⅱ algorithm is used to generate recommendations for target users.The algorithm effectively ensures the diversity of the population and solves the problem of premature convergence of the population.Experimental results show that the algorithm can effectively improve the diversity of recommendation results without reducing the accuracy.