Research on Countermeasures of Network Collaborative Filtering Based on Genetic Algorithm
Download as PDF
DOI: 10.25236/icmmct.2023.028
Author(s)
Xuechun Yang, Xing Ma, Shuai Zheng, Yongmei Bao
Corresponding Author
Yongmei Bao
Abstract
CF(Collaborative Filtering) also faces some problems. Such as cold start problem, scalability problem and data sparsity problem. The cold start problem is divided into system cold start, user cold start and project cold start. By using GA, the differentiation of user similarity is improved, which is beneficial for rating prediction and recommendation. Different from the classic CF algorithm, we use the constructed GA to predict the vacancy value in the scoring matrix before calculating the similarity. The crossover operation in GA selects the parent chromosome based on the crossover probability, and generates new chromosomes through crossover calculation. This paper attempts to study the countermeasures of network CF problem based on GA(Genetic Algorithm). The experimental results show that when the data is extremely sparse, the Mae of various algorithms will increase, but the algorithm in this paper has the minimum Mae value, which keeps around 0.79 on average, reflecting that the algorithm in this paper can better adapt to the data with different sparsity. Through the matrix filled with GA, the similarity of users is calculated, and the nearest neighbor set of the target user is found. In the process of crossover and mutation. The evolution of individuals is easy to produce infeasible individuals. After we get the nearest neighbor set of the target user, we can predict the user's score of unrated items according to the evaluation of the user's nearest neighbors.
Keywords
Genetic algorithm; Network CF; Countermeasure