基于用户特征的馆藏图书智能融合聚类推荐仿真

Simulation of Intelligent Fusion Clustering Recommendation for Library Collections Based on User Characteristics

为了提高馆藏图书的检索效率,提出基于用户特征的馆藏图书智能融合聚类推荐方法。首先,通过TF-IDF方法提取用户对图书资源的偏好特征,将时间系数融入衰减函数中分析用户在不同时间段的偏好特征变化情况,更新用户偏好特征;然后,采用K-means算法根据偏好特征对用户实施聚类处理,利用改进后的人工蜂群算法对聚类中心展开优化,完成用户聚类;最后,根据用户之间的偏好特征相似度获取目标用户的最近邻居,计算目标用户与最近邻居对图书资源的评分相似度。根据计算结果建立加权融合模型,基于预测用户对未阅读图书资源的评级,选择评级高的图书资源生成推荐列表,实现馆藏图书的智能融合聚类推荐。仿真实验结果表明,所提方法的聚类精度高、覆盖率高、推荐效果好、多样性好。

In order to improve the retrieval efficiency of library collections,a user feature-based intelligent fusion clustering recommendation method for library collections is proposed.Firstly,the TF-IDF method is used to extract user preference features for book resources,and the time coefficient is incorporated into the decay function to analyze the changes in user preference features at different time periods.Based on this,the user preference features are updated.Using the K-means algorithm to cluster users based on their preference characteristics,and using the improved artificial bee colony algorithm to optimize the clustering centers and complete user clustering;Finally,based on the similarity of preference features between users,the nearest neighbor of the target user is obtained,and the similarity in ratings of book resources between the target user and the nearest neighbor is calculated.Based on the calculation results,a weighted fusion model is established to predict the user's rating of unread book resources.High rated book resources are selected to generate a recommendation list,achieving intelligent fusion clustering recommendation of library collections.The simulation experiment results show that the proposed method has high clustering accuracy,high coverage,good recommendation effect,and good diversity.