欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
传统的基于项目的协同过滤算法，不能很好地解决数据稀疏和新项目问题（冷启动）带来的推荐质量下降的问题。笔者从智能检索的思想出发，提出一种新的基于知识的协同过滤推荐算法。该算法借助于领域本体，表达语义知识，增加了项目之间的关联信息；考虑到领域本体中结点、边、深度和密度对相似性计算的不同影响，算法结合信息论中的互信息相关概念，对相似性计算公式进行改进，提高了运算精度。实验结果表明，该算法相对于传统的基于项目的协同过滤推荐算法而言，可有效缓解由数据集稀疏和冷启动带来的问题，显著提高推荐系统的推荐质量。
英文摘要：
The traditional item-based collaborative filtering（CF） algorithm often suffers from two important problems： sparsity and new_ iteam （ cold star）, so the performance of recommendation is low. According to the intelligence information retrieval system, the author presents the new algorithm of collaborative filtering based on knowledge. The semantic knowledge expresses the relation between items and adds useful information. Thinking that the effect of the edge, note , depth and density in ontology is different to simility, the algorithm depends on some concept of mutual information in information theory and adjusts the simility formulate to improve the precision. Experimental results show that the algorithm can achieve better prediction accuracy than traditional item-based CF algorithms. Furthermore, the algorithm can alleviate the dataset sparsity and cold star problem.
同期刊论文项目
同项目期刊论文
期刊信息
位置：