Construction of Decision Diagrams for Product Configuration

Knowledge compilation is a well-researched field focused on translating propositional logic formulas into efficient data structures that allow polynomial-time online queries related to the SAT problem. Knowledge compilation techniques can be used to partition product configuration tasks into two distinct phases: fast online processing and slow offline preprocessing. Binary Decision Diagrams (BDDs) are widely studied in this area and provide a graph representation of Boolean formulas. However, BDD construction can be time-consuming, particularly for large instances, as their size grows exponentially with the number of variables. This paper explores methods to improve BDD construction time, including optimizing variable ordering. The evaluation involves applying these techniques to formulas in Rich Conjunctive Normal Form, comparing the results with Sentential Decision Diagrams. The experiments use CAS Software AG benchmarks.