Abstract

Examination Fraud is a menace that could seriously affect the society if unchecked. Widespread examination fraud in the academic sector could lead to the graduation of half-baked graduates, who are unable to prove their worth in the professional world. Such people could infiltrate the leadership cadre of the society, and possibly propagate academic decadence. The conduct of examinations is a pervasive human activity, which is not limited to the academic world. Employment aptitude tests, for instance could be prone to examination fraud. It is therefore necessary to harness the power of Computational Sciences to deal with the problem of examination fraud. This research proposes the use of randomized algorithms to tackle the problem of examination malpractices, especially in the academic institutions. One serious form of examination cheating considered in this work is the one involving the collusion of individuals, who may have had prior knowledge of each other’s academic ability. Such a prior knowledge makes it possible for some candidates to depend on a set of academically top candidates in the examination hall, in order to scale through their examinations. It is also possible for some candidates to have prior knowledge of the examination papers through pre-examination leakage. The focus of the proposed algorithm is to introduce an element of randomization, such that collusion between the candidates or between the candidates and the invigilators could be controlled. The proposed algorithm is made up of three variations, all of which have been presented, and analyzed. The first option is to computationally randomize the seating arrangement, so as to limit the possibility of each student sitting close to a very familiar person. The second option is to introduce randomization into the serial arrangement of the examination papers, such that the candidates sitting near one another may not be able to collude effectively. The third option is to introduce variations into the content of the actual examination, such that some candidates sitting close to themselves may not have the same examination questions. The overall algorithm is fully analyzed, and a proof is given, to show that the model treats all the examination candidates with fairness. The system is designed, and implemented in MATLAB.