中文摘要：

高维、稀疏的用户．项目评分矩阵对基于项目的协同过滤推荐算法造成严峻的可扩展性问题。传统的解决方法是离线计算项目相似性并保存在系统中以供算法调用，但是不能充分利用最新评分数据以体现用户兴趣的变化。针对上述问题，提出了适合在线应用的协同过滤项目相似性增量更新机制，使得推荐系统在当前用户提交项目评分之后，能够实时完成相应项目与其他项目之间的相似性数据更新，从而推荐系统可以基于最新的项目相似性数据进行推荐处理，以适应用户兴趣的变化。实验结果表明，本文提出的项目相似性增量更新机制能够有效提高基于项目的协同过滤算法可扩展性。

英文摘要：

Higher-dimensional, sparse matrix of user-item ratings brings serious scalability problem to item-based collaborative filtering recommendation algorithm. Conventional solution is computing item similarities offline and saving them in system to be read by recommendation algorithm. However, this solution can not reflect user interest changes. To solve the above problem, an incremental updating mechanism of item similarity which suits for online applications is proposed. After the submitting of one new rating by active user, recommender system can finish the real-time updating of item similarity between target item and other items. Hence, recommender system can go on the recommendation processing based on newest item similarity to incorporate user interest changes. The experimental results show that the proposed incremental updating mechanism can efficiently improve the scalability of item-based collaborative filtering.