Abstract: In the age of big data, constantly changing data sourcesmake it a great challenging task to assure data timeliness in enterprise information systems. Taking the dynamic purchase data in an enterprise resource planning(ERP)system as the research object, this paper mainly studies how and when to update the purchase data in ERP systems from the application driven aspect so as to minimize the sum of updating cost and staleness cost. We apply Markov decision process to the problem of purchase data updating in ERP systems and design a dynamic programming algorithm to obtain the optimal updating policy. Compared with the traditional fixed interval updating policies throughnumerical analyses, the proposed application driven purchase data updating policy could not only assure data timeliness but also save enterprise cost effectively.

Key words: application driven, data timeliness, data updating, data quality, ERP

摘要： 大数据时代下,外界数据源的快速变化为保证企业信息系统中动态数据的及时性带来巨大挑战。以企业资源计划(Enterprise Resource Planning, ERP)系统中动态的采购数据为研究对象,从数据应用角度研究如何以及何时更新ERP系统中的采购数据,从而使系统更新成本与数据过时成本最优的问题。将ERP系统中采购数据动态更新过程刻画为马尔科夫决策过程,设计了求解数据最优更新策略的动态规划算法。通过数值分析的对比结果发现,相对于传统的基于固定周期的数据更新策略而言,应用驱动的非固定周期动态数据更新策略不仅能有效保证动态数据的及时性,还能够有效降低企业的数据更新成本。

关键词: 应用驱动, 数据及时性, 数据更新, 数据质量, ERP