摘要
为了解决如何利用无限容量的数字资源与有限的用户信息及时而精准地向用户推荐可用的电子资源等问题,本研究设计了一种可以过滤不良信息的准确推荐算法。该算法为基于协同过滤与内容推荐的混合推荐算法,其中,协同过滤算法提取用户的特征,计算用户间的相似度并对相应的资源进行打分估计从而根据估分进行推荐;而基于内容推荐的算法用于处理用户无法求算相似度的冷启动问题,不良信息利用基于内容推荐的算法提取关键词并与不良关键词库对照,然后从前述推荐结果去掉不良信息;算法还考虑了用户兴趣随时间变化的问题。使用大规模图书馆数字资源数据集对本研究算法进行测试,结果表明,使用本研究算法,邻居数的增加对推荐精度有改善作用;对使用平均相似度和加权相似度的结果比较表明,加权相似度可以获得更好的推荐效果;加入时间因素,可以有效改进推荐精度,进而实现了对不良信息的过滤,保证了资源的质量。本研究算法基本实现了精准推荐,可适用于大数据环境下数字资源的推荐操作。
At present,how to utilize the unlimited capacity of digital resources and limited user information to recommend available e-resources to users in a timely and accurate manner is the difficulty faced by digital resource management.To address this difficulty,in this study,a recommendation algorithm was presented which can provide the user with what the user really want or find.This algorithm is a hybrid method of collaborative filtering method and content based method,of which the former calculates the similarity of the users to predict the score,and then the score will be acted as the basis to recommendations.The latter targets to solve the recommendation problem without enough information to calculate the similarity.Additionally,this algorithm will take time stamp into consideration.A large scale digital resource data set from library was employed to test our algorithm,the results indicated several points:this algorithm can be improved by increasing the neighbor number;the weighted similarity algorithm can outperform the average similarity algorithm;by taking the time stamp in consideration,the algorithm can be improved.This method can make recommendation with a high accuracy,and can be employed for digital resource recommendation in big data context.
作者
郭笃凌
闫长青
GUO Du-ing;YAN Chang-qing(Department of Public Course Teaching,Shandong University of Science and Technology,Taian 271019,China;College of Intelligent Equipment,Shandong University of Science and Technology,Taian 271019,China)
出处
《印刷与数字媒体技术研究》
CAS
北大核心
2024年第1期113-121,共9页
Printing and Digital Media Technology Study
关键词
数字资源
推荐系统
相似性度量
混合推荐算法
Digital resource
Recommendation system
Similarity algorithm
Hybrid recommendation algorithm
