Feedback plays important role in education, especially in student- centred learning. 21st century’s skills such as system thinking and non-routine problem solving are complicated to assess, because they are based on structural knowledge which is more than memorising of pure facts and definitions. A computer-based concept mapping is one of suitable tools to support assessment of structural knowledge. Algorithms to extend functionality of the intelligent concept map based knowledge assessment system IKAS to turning it into a knowledge self-assessment system are proposed. This extension involves the use of learning objects for correction of student’s knowledge based on analysis of his/her errors in concept mapping tasks. A personalized learning path is generated depending on identified errors. Learning objects are created in a new manner, i.e., each learning object is connected to at least two interrelated concepts, thus a structure of concept relationships is better delivered to a student.