Information Processing for Generating Recommendations Ahead of Time in an IoT-Based Environment

Introduction

The evolution of the Internet of Things (IoT) has significantly contributed to the development of the sensing enterprise concept, which deals with the use of multi-dimensional data captured through physical and virtual sensors generating events and providing added value information. This huge amount of real-time data leads to the need for sensor-based, real-time data-driven information systems incorporating efficient processing technologies and mechanisms in order to provide meaningful insights about potential problems (Bousdekis et al., 2015). Event monitoring and big data processing are of outmost importance because they enable not only observing current problems, but also identifying that the problem may appear, leading to the possibility to decide and act ahead of time, in a proactive way (Engel et al., 2012; Bousdekis et al., 2015). In this sense, proactive decision making (Engel et al., 2012) can be further developed and validated in the frame of a proactive Event Driven Architecture (EDA) integrating real-time data from various sources (e.g. sensors), facilitating large-scale and real-time processing of these data and combining historical data and domain knowledge with current data streams in order to provide recommendations about which actions to implement and when. Recommendations are provided on the basis of predictions with the aim to optimize the business utility function. Predictions are triggered when an unusual situation is detected based on the sensor-based real-time observations.

Despite the increasing amount of research regarding detection and prediction methodologies, algorithms and information systems, automation of decisions ahead of time is still an unexplored area. In this paper, the authors describe the design requirements for the implementation of a system for context-aware proactive decision making in sensing enterprises. More specifically, they describe how the information is processed so that decision making ahead of time is facilitated. In this sense, this paper does not focus on the algorithms implemented, but on information processing among the constituting blocks of the system architecture in order to generate recommendations ahead of time in an IoT-based environment. The current paper constitutes a significantly extended version of (Bousdekis et al., 2017a). The rest of the paper is organized as follows. First, there is a review of background and works related to proactive event-driven computing. Then, it outlines the conceptual architecture for continuously improved context-aware proactive event-driven decision making in an IoT-based environment. The system design requirements for deciding and acting ahead of time are also presented, while a system walkthrough is provided. Finally, the system evaluation results are highlighted, and the conclusions are discussed.