融合多信息的跨域推荐算法

A Cross-Domain Recommendation Algorithm Integrating Multi-Information

针对目前跨域协同过滤算法仅通过评分矩阵相似性进行信息提取推荐,并未利用数据中含有的更多信息,从而导致推荐结果不理想的情况。提出一种融合多信息的改进跨域协同过滤算法。算法通过改进传统跨域协同过滤中的信息提取方式,融入了数据源中的时间与类型信息,提高了信息提取的精度与推荐的准确性。通过在MovieLens数据集与豆瓣数据集上进行对比实验,结果表明,跨域推荐算法能够在多域间进行信息传递,融入了多信息的跨域推荐算法能更为有效地提升推荐的准确性。

For the current cross-domain collaborative filtering algorithm,only information extraction and recommendation is performed through the similarity of the score matrix,and more information contained in the data is not utilized,resulting in unsatis⁃factory recommendation results.Propose an improved cross-domain recommendation algorithm integrating multi-information.By improving the information extraction method in traditional cross-domain collaborative filtering,the algorithm integrates the time and type information in the data source,and improves the accuracy of information extraction and recommendation.Through com⁃parative experiments on the MovieLens dataset and Douban dataset,the results show that the cross-domain recommendation algo⁃rithm can transfer information between multiple domains,and the cross-domain recommendation algorithm incorporating multiinformation can more effectively improve the accuracy of the recommendation.