Motorola provides a database that contains time-sequenced information collected from a set of semiconductor (wafer) manufacturing processes. In a typical scenario in this database, a wafer can spend as much as 200 seconds in a machine and the machine is controlled by more than 100 parameters. So each data object has a maximum of 20,000 measurements.

Given a set of positive and negative wafer operation examples, the task is to induce a set of probabilistic rules that can detect defective wafers during their manufacturing process in real-time. Such detection may occur even before the manufacturing process is completed so that necessary corrections might be injected into the process, to repair the product before it is too late.

ECC provides a database that contains results representing over 30 years of chemical research and experiments in ECC, in addition to information about the ingredients and properties of various compounds. The database has more than 50 tables, and largest table has over 500,000 records.

The task is build chemical predictive models. These models will be used by chemists to make predictions either from properties to compositions, or form compositions to properties. The former can facilitate chemists to design new polymers to meet customer's needs, and the later can help chemists to gain valuable knowledge from large amount of experimental data.