Automatically prioritizing software development tasks extracted from codes could provide significant technical and organizational advantages. Tools exist for the automatic extraction of tasks, but they still lack the ability to capture their mutual dependencies; hence, the capability to prioritize them. Solving this important puzzle is the goal of the presented industrial challenge.

Ample research has been carried out in the area of collusion, plagiarism and e-learning. Collusion is a form of active cheating where two or more parties secretly or illegally corporate. Collusion is at the root of common knowledge plagiarism. While plagiarism requires two or more entities to compare, collusion can be determined in isolation. It is also possible that collusion do not lead to positive plagiarism checks. It is therefore the aims of this preliminary study to: (i) identify the factors responsible for collusion in e-learning (ii) determine the prominent factor that is representative of collusion and (iii) through user behaviour including, but not limited to, application switching time, determine collusion. We claim that user computer activities and application processes can help understand user behaviour during assessment task. It is on this premise that we develop a machine learning model to predict collusion through user behaviour during assessment task