针对传统协同过滤算法依赖单一用户需求形态影响推荐效果的问题,提出一种基于用户多态聚类的数字图书馆个性化推荐方法。该方法以改进的海明距离计算候选邻居集,结合多态相似度进行二次聚类,预测用户的多态需求度并形成推荐。实验表明,使用多态聚类产生的推荐精确度上优于单一聚类产生的推荐。 Aiming at the problem that the traditional collaborative filtering algorithm relies on the effect of single user’s requirement on the recommendation effect, a personalized recommendation method of digital library based on user’s polymorphic clustering is proposed. This method uses the improved Hamming distance to calculate the candidate neighbors, and then carries out the second clustering based on the polymorphism similarity to predict the user’s polymorphic demand and form a recommendation. Experiments have shown that the use of polymorphic clustering produces recommendations that outperform recommendations made by a single cluster.