摘要
随着社会潮流趋势的演进,如何更好地满足个性化需求已成为个性化推荐服务研究领域的新趋势,推荐系统对帮助用户在海量的线上异构数据中快速发现其感兴趣的内容具有广泛的应用。为了在复杂场景中有效缓解推荐系统研究领域普遍存在的数据稀疏和冷启动等问题,同时在复杂环境下提高其推荐项目的准确性和多样性,提出一种在矩阵分解技术的基础上同时考虑社交网络推荐的新方法。首先,通过将粒子群优化技术(PSO)与K-harmonic means(KHM)聚类算法融合,设计了一种混合聚类算法并对用户进行聚类,然后在相似度计算模型中引入多因素,利用矩阵分解技术计算用户的行为偏好,最终获取用户的最佳项目推荐列表方案。研究对Book-Crossing书评基准数据集进行仿真分析,结果表明提出的新方法具有较好的推荐准确性和多样性。
With the evolution of social trends,how to better meet the personalized requirements has become a new trend in the research field of the personalized recommendation service,and recommendation system in the line of huge amounts of heterogeneous data to help users quickly find their interest in the content have a wide range of applications. To relieve recommendation system research field in complex scene common problems such as data sparseness and cold start,at the same time in a complex environment to improve the recommendation accuracy and diversity of the project,this paper puts forward a matrix decomposition technique on the basis of the new method considering the recommendation of social network. First,by putting the particle swarm optimization（ PSO） and K-harmonic means（ KHM） clustering algorithm,we design a hybrid clustering algorithm and the user clustering, and then the similarity calculation model introduced in many factors,matrix decomposition technique is used to calculate the behavior of user preferences,and ultimately achieve the best projects recommended list of the user program. The results show that the proposed method has better recommendation accuracy and diversity.
出处
《图书馆学研究》
CSSCI
北大核心
2018年第14期31-37,共7页
Research on Library Science
基金
国家自然科学基金"国家自然科学基金资助计划历史数据分析及信息系统功能需求设计"(项目编号:J1624001)的研究成果之一
北京物资学院研究生科研创新项目
