摘要
基于标签的推荐算法在景点推荐领域取得了良好效果,但仍然存在一些问题,如仅采用用户对景点的评分值表示用户对标签的喜爱程度,忽略了用户、标签、景点之间的关联,从而导致结果精度不高。提出一种融入景点标签的矩阵分解个性化推荐方法,通过文本挖掘技术构建适用于景点推荐领域的景点标签,并将其引入矩阵分解推荐算法的因子向量,然后利用矩阵分解技术深入挖掘用户、标签、景点之间的潜在联系,从而预测用户对景点标签的感兴趣程度,最终通过用户对景点标签的感兴趣程度以及景点对标签的隶属度预测用户对景点的评分值。实验结果表明,该方法相比基于用户的协同过滤景点推荐算法,MAE和RMSE分别降低68.28%、61.23%,相比基于标签的协同过滤景点推荐算法,MAE和RMSE分别降低67.02%、59.93%,其性能明显优于现有相关景点推荐算法,能够为景点推荐提供有力支撑。
The tag-based recommendation algorithm has achieved good results in the field of attraction recommendation,but there are still some problems.For example,only the user’s rating value of the attraction is used to indicate the user’s preference for tags,ignor⁃ing the relationship between users,tags,and attractions.As a result,the accuracy of the result is not high.This paper proposed a person⁃alized recommendation method of matrix factorization fused with attractions tags.This method uses text mining technology to construct scenic spot tags suitable for the field of scenic spot recommendation,and introduces them into the factor vector of the matrix factoriza⁃tion recommendation algorithm,and then uses matrix factorization technology to deeply explore the potential connections between us⁃ers,tags,and scenic spots to predict users.The degree of interest in the tag of the scenic spot is finally predicted by the user’s degree of interest in the tag of the scenic spot and the degree of membership of the scenic spot to the tag.The experimental results show that the method proposed in this paper reduces the MAE and RMSE of the user-based collaborative filtering attraction recommendation algo⁃rithm by 68.28%and 61.23%,respectively,and reduces the MAE and RMSE respectively compared to the tag-based collaborative fil⁃tering attraction recommendation algorithm 67.02%and 59.93%,its performance is significantly better than the existing related attrac⁃tions recommendation algorithm,which can provide strong support for attraction recommendation.
作者
张鑫
许璐璐
ZHANG Xin;XU Lu-lu(College of Geodesy and Geomatics,Shandong University of Science and Technology,Qingdao 266590,China;Yantai Yufeng Geological Technology Service Co.,Ltd.,Yantai 264010,China)
出处
《软件导刊》
2021年第4期199-204,共6页
Software Guide
关键词
矩阵分解
景点标签
协同过滤
个性化推荐
文本挖掘
matrix decomposition
attraction label
collaborative filtering
personalized recommendation
text mining
