Computational scientific discovery is concerned with applying computational methods to automate scientific and engineering activities, such as establishing models of real-world systems from observational data. Experience with applying computational discovery methods to real-world problems shows that the process of knowledge discovery and modeling in science is often an incremental activity: scientists usually build on existing theories and models. While most of the traditional methods for computational discovery do not address the incremental nature of scientific activities, inductive database can provide scientists with an interactive environment for assisting and properly documenting the discovery process. In the talk, we will discuss this and other benefits of using inductive databases for computational scientific discovery.