Over the last couple of decades the technological advancement had a huge impact on the research of intelligence as well. More and more of the traditional tests have been digitialized. Certain steps of the test creation and evaluation process became automatizable. One of the most popular methods of measuring intelligence is the figural abstraction test. The first implementation of this method (RPM – Raven’s Progressive Matrices) has been actively used since its creation. A vast amount of different variation has been since implemented, most of them being accessible online. The mobile research group of our department has developed a framework for educational applications, which provides the opportunity for these to hand out tasks to the users with adaptively changing difficulty levels. For this adaptiveness it is necessary to have different elements for feedback, such as the current performance of the user or even physiological signs. Here arised the need for a self-standardizing intelligence test which is capable of determining the competence levels of the users, providing a starting point for adaptiveness.
Last year a smart-phone version of a figural abstraction test has been developed in which the exercises were randomly generated using pre-determined patterns. This kind of task creation resulted in a vast amount of different exercises compared to traditional tests. The next step, of which this paper is about, is to automatically generate these patterns, which could result in an even higher number of exercises. In this paper a method like this will be presented along with the planning and implementation of a system using it.