An Application to Discover Cheating in Digital Exams

Abstract

Cheating is a common problem for both, paper-based and electronic examinations. Therefore, it is desirable to be able to detect cheating reliably. Since it is not always possible to recognize a cheating attempt in situ, other ways to detect cheating have to be found. One way is to analyze the answers that students hand in to verify that a particular student is in fact the author of those answers. This can be done based on the assumption that students develop an individual style for answering certain types of assignments, which can be extracted using techniques of artificial intelligence and then compared to reference material for which the author is verified. This paper presents FLEXauth, an application which tackles this task for electronic programming exams with Machine Learning techniques and discusses the state of the art of author verification as well as first results and open research questions that have to be addressed for the further development of FLEXauth.

Ähnliche Arbeiten