基于物品的协同过滤算法对“宁波地铁go”用户个性化推荐系统研究

Research on Personalized Recommendation System of “Ningbo Metro Go” Users Based on Item-based Collaborative Filtering Algorithm

随着数据挖掘技术的不断发展,个性化推荐系统在各个领域已被广泛应用,电子商务平台根据用户的历史数据,向其推荐感兴趣的商品,但如今也存在着推荐商品精确度低,特征提取能力有限,标签本身存在冗余问题等。文章通过基于物品的协同过滤算法,在用户群中找到指定用户的相似邻居用户,综合这些邻居用户的行为数据,再引入物品售价权重这一隐氏特征,避免热门物品对推荐结果的恶意干扰,改进传统算法,从而推荐更合适的商品。实验表明该算法对推荐预测有较好的效果。

With the continuous development of data mining technology,personalized recommendation system has been widely used in various fields.The E-commerce platform recommends products of interest according to the user's historical data,but now there are also problems of low accuracy in recommending commodities,the feature extraction capability is limited,and the tag itself has redundancy problems.Through the item-based collaborative filtering algorithm.This paper finds similar neighbor users of the specified users in the user group,integrates the behavior data of these neighbor users,and then introduces the hidden feature of the item price weight to avoid malicious interference of the popular items on the recommendation results.Improve traditional algorithms to recommend more suitable products.Experiment shows that the algorithm has a good effect on the recommended prediction.