基于SGA-RBF的协同过滤算法研究

Research on Collaborative Filtering Algorithm Based on SGA-RBF

为减小协同过滤算法造成的误差,提高推荐的效率和质量。文章用遗传算法优化RBF神经网络的初始权值,提出了SGA-RBF神经网络模型,在项目相似度的基础上,将SGA-RBF神经网络与协同过滤算法结合,预测了未评分项目的分数,将预测评分和实际评分进行比较,并计算了平均相对误差。经过遗传算法的优化,RBF神经网络的初始权值更加准确。实验结果显示,改进的协同过滤算法使预测更加精确,MAE值更低,具有一定的现实价值。

In order to reduce the error caused by collaborative filtering algorithm and improve the efficiency and quality of recommendation, this paper optimizes the initial weight of RBF neural network by using genetic algorithm, and proposes a SGA-RBF neural network model. And then based on the similarity of items, the paper combines the SGA-RBF neural network with the collaborative filtering algorithm to predict the scores of the ungraded items, to compare the predicted scores with the actual scores, and calculate the average relative error. After the optimization of genetic algorithm, the initial weight of RBF neural network is more accurate. The experimental results show that the improved collaborative filtering algorithm makes the prediction more accurate and the MAE value lower, so it has certain practical value.