Implementing Smart Processes using IoT, Machine Learning, and Business Rules

This unique book proposes a systematic methodology spanning the full lifecycle of such so-called smart process applications (SPAs) and illustrates it with two realistic case studies. SPAs are fundamentally different from traditional transaction-driven IT applications, from functional and architectural requirements points of view. They silently monitor independently running processes and only spring into action when they recognize some situations, to suggest or execute actions that will help the process under observation to achieve or prevent certain outcomes. With SPAs, the IoT-enabled context awareness, cloud-enabled distribution, and AI-enabled decision making—a combination of machine learning and business rules—are a matter of domain necessity and not mere technical wizardry. At the core of the proposed development methodology is a requirements framework that helps translate general domain requirements—staying healthy, preventing accidents, etc.—into precise functional and architectural requirements. Topics and features: · An integrated development methodology that spans the entire lifecycle of SPAs from domain requirements to architecture, detailed design, coding, testing, and maintenance · A pragmatic approach to hybrid AI, advocating the use of machine learning or business rules depending on the task at hand, the state of knowledge, and the availability of historical data · A combination of state-of-the-art developments with lessons learned from dozens of consulting engagements in industry · Two running case studies help explore the design space for the component technologies, taken separately, and in an integrated fashion · A companion evolving GitHub repository that hosts various development deliverables for the case studies Practicing software engineers who work with the domain and architectural dimensions of SPAs will find food for thought and ready-to-use generic, and technology-specific, design patterns. Academic researchers interested in either method engineering or in developing domain-specific SPAs will find great value in this book. Hafedh Mili is full professor of Computer Science at the Université du Québec à Montréal, Canada. Ghizlane El Boussaidi is a full professor affiliated with the Dept. of Software and IT engineering at École de Technologie Supérieure, Canada.