Symbolic Artificial Intelligence Methods for Prescriptive Analytics

Prescriptive analytics in supply chain management and manufacturing addresses the question of “what” should happen “when”, where good recommendations require the solving of decision and optimization problems in all stages of the product life cycle at all decision levels. Artificial intelligence (AI) provides general methods and tools for the automated solving of such problems. We start our contribution with a discussion of the relation between AI and analytics techniques. As many decision and optimization problems are computationally complex, we present the challenges and approaches for solving such hard problems by AI methods and tools. As a running example for the introduction of general problem-solving frameworks, we employ production planning and scheduling. First, we present the fundamental modeling and problem-solving concepts of constraint programming (CP), which has a long and successful history in solving practical planning and scheduling tasks. Second, we describe highly expressive methods for problem representation and solving based on answer set programming (ASP), which is a variant of logic programming. Finally, as the application of exact algorithms can be prohibitive for very large problem instances, we discuss some methods from the area of local search aiming at near-optimal solutions. Besides the introduction of basic principles, we point out available tools and practical showcases.