MapReduce环境下面向用户偏好的top-k连接查询处理方法

发布时间：2018-03-24 18:39

本文选题：MapReduce　切入点：top-k连接　出处：《小型微型计算机系统》2017年11期

【摘要】：随着互联网技术的迅猛发展和大数据时代的来临,面向海量数据的top-k连接查询返回组合得分最大(或最小)的k个连接结果,能够广泛应用到商业模式预测、用户需求分析、商品营销统计等领域.然而,在实际应用中,不同的用户对参与连接的数据表中的某些维度往往会有不同的要求,进而产生了不同的用户偏好.为了选出符合用户偏好的连接结果,需要针对用户偏好对连接后的结果进行筛选.因此,本文针对不同用户在查询时有各自不同偏好的问题,提出了在MapReduce环境下基于用户偏好的top-k连接查询处理方法.首先,在预处理阶段将两表进行连接操作;其次,提出利用skyline技术求解用户偏好的方法,通过分区间支配关系及虚拟最小点的判定,过滤不满足用户偏好的元组;最后,再通过打分函数找到需要的top-k连接查询结果.大量实验表明,本文提出的基于用户偏好的top-k连接查询处理方法能够有效满足用户偏好,具有良好的有效性和扩展性.
[Abstract]:With the rapid development of Internet technology and the advent of big data era, top-k join queries for mass data return k join results with the largest (or least) score, which can be widely applied to business model prediction and user demand analysis. However, in practical applications, different users often have different requirements for certain dimensions in the connected data table. In order to select the join result that accords with the user preference, we need to filter the join result according to the user preference. Therefore, this paper aims at the problem that different users have different preferences when querying. This paper proposes a top-k join query processing method based on user preference in MapReduce environment. Firstly, two tables are joined in the preprocessing stage. Secondly, a method to solve user preference using skyline technology is proposed. By determining the subdivision dominance relation and the virtual minimum point, the tuple that does not satisfy the user's preference is filtered. Finally, the required top-k join query results are found by scoring function. A large number of experiments show that, The proposed top-k join query processing method based on user preference can satisfy user preference effectively and has good efficiency and expansibility.
【作者单位】：辽宁大学信息学院;
【基金】：国家自然科学基金项目(61472169,61502215)资助国家重点研发计划项目(2016YFC0801406)资助辽宁省教育厅科学研究一般项目(L2015193)资助辽宁省博士科研启动基金项目(201501127)资助
【分类号】：TP311.13

【相似文献】