融合情景和社会网络的数字资源服务推荐算法  

Recommendation algorithms for digital resource services integrating context and social network

作　　者：张龙昌 白静 ZHANG Longchang;BAI Jing(School of Information Engineering,Suqian University,Suqian 223800,China;School of Management Science and Engineering,Dongbei University of Finance and Economics,Dalian 116025,China)

机构地区：[1]宿迁学院信息工程学院,江苏宿迁223800 [2]东北财经大学管理科学与工程学院,辽宁大连116025

出　　处：《江苏海洋大学学报(自然科学版)》2025年第1期62-70,共9页Journal of Jiangsu Ocean University:Natural Science Edition

基　　金：国家社会科学基金资助项目(19BTQ028)。

摘　　要：异构环境下,为进一步提高基于用户的数字资源服务协同过滤推荐精准度和缓解冷启动、数据稀疏对推荐质量的影响,提出融合读者情景和社会网络的数字资源服务协同过滤推荐方法。该方案设计融合评分相似度计算、情景相似度计算、邻居信任度计算的读者近邻相似度综合计算方法,提出基于读者近邻相似度综合计算的评分预测和服务推荐方法,基于迪杰斯特拉思想设计了信任度计算算法,求解读者邻居信任度。通过在真实数据集上验证,所提出的方法在一定程度上改善了基于用户协同过滤推荐的精准度,并且一定程度上缓解了冷启动、数据稀疏对推荐质量的影响。In the heterogeneous environment,to enhance the accuracy of user-based collaborative filtering recommendation for digital resource services and mitigate the impact of cold start and data sparsity on recommendation quality,this study proposes a digital resource service collaborative filtering recommendation method that integrates reader context and social network.The proposed solution devises a comprehensive calculation method for determining reader neighbor similarity by integrating score similarity calculation,context similarity calculation,and neighbor trust degree calculation.Furthermore,it presents a rating prediction and service recommendation method based on the comprehensive calculation of reader neighbor similarity,drawing inspiration from Dijkstra,an algorithm is devised to calculate trust degrees in order to ascertain the reliability of reader neighbors.Experimental results obtained from real datasets validate that the proposed approach moderately enhances the accuracy of user-based collaborative filtering recommendations while also partially alleviating the impact of cold start and data sparsity on recommendation quality.

关 键 词：情景 社会网络 数字资源服务 协同过滤 相似度计算 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]