通过分析在用户评分数据极端稀疏的情况下,现有的基于项目评分预测的协同过滤推荐算法中项目之间的相似性度量不准确以及新项目的冷开始问题,提出了一种优化的基于项目评分预测的协同过滤推荐算法。该算法在计算项目之间的相似性时,既考虑了项目的评分相似性,又考虑了项目的特征属性相似性。实验表明,优化后的算法使计算出的项目之间的相似性更准确,并有效地解决了新项目的推荐问题,使得数据稀疏性对推荐结果的负面影响变小,显著提高了系统的推荐质量。
By analyzing the inaccuracy of item similarity and new item recommendation in present collaborative filtering algorithm based on item rating prediction under data sparsity condition, this paper proposed an optimized collaborative filtering recommendation algorithm based on item rating prediction. This algorithm considered for user rating and item attribute item similarity calculation. The experiment shows that the optimized algorithm makes the similarity bet,seen items more accurate and solves the problem of new item recommendation effectively. The algorithm reduces the negative effect on the final recommendation and can provide better recommendation results for the system.