Abstract:
Many applications in diverse areas such as manufacturing systems and sustainable energy systems require making complex decision based on stochastic data. Probabilistic and statistical databases are excellent for declarative queries, but do not support more complex stochastic processes defined through stochastic simulation. Stochastic simulation allows the user to create complex models to simulate an event, but does not support declarative formulation of queries and is time consuming. To support this type of application, this dissertation introduces Simulation Query Language, SimQL, a language extends the database query language SQL with stochastic attributes/ random variables defined by simulation. I also propose algorithms for computing top-k answers based on partial search space exploration for continuous decision variables using regression analysis and algorithms based on enumeration heuristics for scheduling problems. I also conducted experimentation comparing the algorithms for continuous decision variables and a case study for a class of scheduling problems.
