大规模在线课程非专用资源预测与查询调度  

Prediction and adaptive query scheduling of non-dedicated resources on massive open online courses

作　　者：侯勇[1] 吾守尔.斯拉木 吐尔根.伊布拉音 周艳慧[2] 张凯[2] 

机构地区：[1]新疆大学信息科学与工程学院,乌鲁木齐830046 [2]国家广播电影电视总局机关服务中心,北京100866

出　　处：《计算机工程与应用》2009年第36期244-248,共5页Computer Engineering and Applications

基　　金：新疆大学青年科研启动基金项目No.070309~~

摘　　要：大规模在线课程系统需要大量资源作支撑。资源使用的动态性及租用的固定性导致资源被浪费。将"非专用"思想应用于系统,可整体降低资源租用量。首先通过马尔可夫"生灭"过程预测资源状态,筛选出"非专用"资源;接着建立查询期望代价矩阵,使用A-MM(Adaptive Min-Min and Max-Min)算法进行自适应查询处理;最后实验表明,预测错误率低,性能稳定;A-MM有较好执行效率和平衡负载能力。Massive Open Online Course system,MOOC,requires a large number of resources for supporting users to get good services.Resources of MOOC used dynamically and fixed time for rental waste a lot of resources.The method of non-dedicated resources is applied to MOOC,to reduce the overall amount of resources to hire.First of all,using Markov birth-death process to predicate resources for getting non-dedicated resources;secondly,setting up query expected cost matrix,QEC;thirdly,using Adaptive Min-Min and Max-Min,A-MM,algorithm for adaptive query scheduling;finally,experiments are done and show that the prediction error rate is low and the performance of prediction is stable;and A-MM has a good efficiency and balance capacity.

关 键 词：非专用资源 自适应查询调度 资源状态预测 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]