Abstract:

Entrance examination for admission to a university is a key aspect of selection of talented and appropriately suited candidates for the study programmes. To achieve this goal, the entrance examinations (either in the form of tests, interviews or their combination) need to be designed in such a way that allows for the expression of the requirements on the candidates as well as for the definition of several potentially well suited types/categories of candidates. The Standard entrance examination tests contain set of various questions, which are selected to exhaustively test the knowledge of each applicant in the relevant areas and map the depth and width of knowledge in the given subject area. Problem solving, aptitude testing as well as motivation assessment, are also sometimes included in entrance examinations. Each part of the entrance examination as well as each question or problem solved during the examination can be of different significance and specific combinations of answers and correct solutions can signify the required combination of skills, knowledge and talent for the given study programme. Usually, the performance during the entrance exam is translated into a score, which is then used to determine the set of applicants to be admitted.

In this paper a new method for entrance examination evaluation, which is based on the fuzzy logic approach, is presented. The set of test questions is divided into subgroups with respect to the focus of each question. Each subgroup is then evaluated individually and these partial evaluations are then aggregated using a fuzzy rule base. The fuzzy rule base contains a set of fuzzy rules (expressed linguistically) that explain how the partial evaluations of each subgroup should be aggregated. This way, more complex aggregation of partial evaluations is possible which reflects the existence of several different possible patterns signifying a desired entrance examination performance.

Keywords:

Entrance examination, evaluation, fuzzy sets, fuzzy rule base, aggregation.