It is difficult if not impossible to derive a model to adequately describe the entire visual, cognitive and preference decision process of image quality evaluation and to replace it with objective alternatives. Even if some parts of the process can be modeled based on the current knowledge of the visual system, there is often a lack of sufficient data to support the modeling process. On the other hand, image quality evaluation is constantly required for those working on imaging devices and software. Measurements and surveys are regularly conducted to test a newer processing algorithm or a methodology. Large scale subjective measurement or surveys are often conducted before a product is released. Here we propose to combine the two processes and apply data mining techniques to achieve both goals of routine subjective testing and modeling. Specifically, we propose to use relational databases to log and store regular evaluation processes. When combined with web applications, the relational databases approach allow one to maximally improve the efficiency of designing, conducting, analyzing, and reporting test data. The collection of large amounts of data makes it possible to apply data mining techniques to discover knowledge and patterns in the data. Here we report one such system for printing quality evaluation and some theories on data mining including data visualization, observer mining, text comment mining, test case mining, model mining. We also present some preliminary results based on some of these techniques.