Classical and Fuzzy Logic Evaluation of Students' Master Theses in Matlab Fuzzy Logic Toolbox Software: Dealing with Subjectivity in Human Reasoning

The two−level evaluation of defined objectives, presented materials and methods and interpretation of results in master theses was done, in order to estimate their scientific contribution and statistical relevance. First level of evaluation was performed using classical methods and consisted of three steps: defining criteria of evaluation, analyzing their fulfilment and positioning 26 master theses into the Likert−type scale in the range from 0 to 1. Second level of evaluation was based on fuzzy logic methodology, conducted mostly in Matlab Fuzzy Logic Toolbox software and consisted of definition of variables, fuzzification, fuzzy inference, defuzzification and interpretation. Obtained marks from two levels of evaluation were than compared. Results indicate that fulfilment of defined criteria of evaluation is moderate. Common mistakes made by authors are accentuated, and clear advices for improving scientific contribution of theses were pointed out here. Classical evaluation marks were higher in 96.15% cases (or 25 out of 26 theses). However, fuzzy approach has advantages, which is also discussed. The interpretation of research results, defined as logical−mathematical argumentation, was found to have the leading role in forming mark in both levels of evaluation.