播存网络环境下UCL推荐多样性优化算法

发布时间：2019-05-10 02:59

【摘要】：播存网络将广播分发模式引入现有互联网体系结构,极大地降低网络共享过程中产生的冗余流量,可有效缓解信息过载问题.播存网络采用统一内容标签(uniform content label,UCL)适配用户兴趣和推荐信息资源,在UCL个性化推荐过程中,如何结合播存网络的富语义、高时效特征,有效地提高UCL推荐列表的多样性,成为播存网络中一个亟需解决的关键问题.针对播存网络环境的需求,提出了一种基于语义覆盖树的UCL推荐多样性优化算法UDSCT,将该问题分为UCL语义覆盖树构建和多样化UCL列表查询2个步骤.在UCL语义覆盖树构建阶段,基于语义覆盖树的若干约束条件,充分考虑UCL语义信息及非语义用户评分信息,同时,较新的UCL具有较高的优先权,以保证列表的时效性;在多样化UCL列表查询阶段,采用简单树查询及启发式列表补充操作,可快速高效地获得多样性优化后的UCL推荐列表,并可进一步根据用户请求快速返回指定的UCL集合.通过理论分析及一系列仿真实验验证,结果证明:UDSCT算法相对于基准算法能够获得更好的多样性优化效果及效率,可有效满足播存网络环境的需求.
[Abstract]:Broadcast network introduces broadcast distribution mode into the existing Internet architecture, which can greatly reduce the redundant traffic produced in the process of network sharing, and can effectively alleviate the problem of information overload. The broadcast network adopts the unified content label (uniform content label,UCL) to adapt the user interest and recommendation information resources. in the process of UCL personalized recommendation, how to combine the rich semantics and high timeliness characteristics of the broadcast network. Effectively improving the diversity of UCL recommendation list has become a key problem to be solved in broadcast network. In order to meet the requirements of broadcast network environment, a UCL recommendation diversity optimization algorithm UDSCT, based on semantic coverage tree is proposed. The problem is divided into two steps: UCL semantic coverage tree construction and diversified UCL list query. In the construction stage of UCL semantic coverage tree, based on some constraints of semantic coverage tree, UCL semantic information and non-semantic user rating information are fully considered. at the same time, the newer UCL has higher priority to ensure the timeliness of the list. In the stage of diversified UCL list query, simple tree query and heuristic list supplement operation can be used to obtain the UCL recommendation list after diversity optimization quickly and efficiently, and the specified UCL set can be returned quickly according to the user request. Through theoretical analysis and a series of simulation experiments, the results show that UDSCT algorithm can achieve better diversity optimization effect and efficiency than benchmark algorithm, and can effectively meet the needs of broadcast network environment.
【作者单位】：东南大学计算机科学与工程学院;计算机网络和信息集成教育部重点实验室(东南大学);
【基金】：国家自然科学基金项目(61472080,61672155) 中国工程院咨询研究项目(2015-XY-04) 国家“八六三”高技术研究发展计划基金项目(2013AA013503) 软件新技术与产业化协同创新中心项目~~
【分类号】：TP391.3

【参考文献】