Approaches to generation of individualized assignments, based on various diversification models, are considered. The techniques are proposed and discussed for constructing a sufficient number of assignment variants of the same complexity, aimed at increasing their diversity and decreasing the possibility of cheating. The built-in automatic assessment function clears the instructor of the tedious work of looking through a large number of similar assignment papers. On the other hand, automatic self-assessment increases the efficiency of students' home work, reduces the turn-around time, i.e. the duration of the cycle "assessment-correction" after the initial completion of the assignment paper by the student.