A Markov-Based Update Policy for Constantly Changing Database Systems

Abstract

In order to maximize the value of an organization's data assets, it is important to keep data in its databases upto-date. In the era of big data, however, constantly changing data sources make it a challenging task to assure data timeliness in enterprise systems. For instance, due to the high frequency of purchase transactions, purchase data stored in an enterprise resource planning system can easily become outdated, affecting the accuracy of inventory data and the quality of inventory replenishment decisions. Despite the importance of data timeliness, updating a database as soon as new data arrives is typically not optimal because of high update cost. Therefore, a critical problem in this context is to determine the optimal update policy for database systems. In this study, we develop a Markov decision process model, solved via dynamic programming, to derive the optimal update policy that minimizes the sum of data staleness cost and update cost. Based on realworld enterprise data, we conduct experiments to evaluate the performance of the proposed update policy in relation to benchmark policies analyzed in the prior literature. The experimental results show that the proposed update policy outperforms fixed interval update policies and can lead to significant cost savings.

ManuKnowVis: How to Support Different User Groups in Contextualizing and Leveraging Knowledge Repositories

“…This knowledge, however, is provided by various stakeholder groups inside organizations, for example about the machine and product properties or data-driven analyses [13]. Hence, the efficient management of these distributed knowledge repositories is a key resource for the successful digitization of manufacturing processes [15], [59].…”

Knowledge Management in the Fourth Industrial Revolution: Mapping the Literature and Scoping Future Avenues

“…For example, in a constantly-connected environment, machines monitor processes continuously and produce reports, increasing the potential for knowledge creation exponentially [9]. As a result of this activity, substantially more unstructured data and information are produced increases, which may congest the information system [10]. For this reason, to truly benefit from this digital transformation, organizations should improve their KM approaches so they can scan and detect meaningful pieces of information and develop more sophisticated uses of this knowledge.…”