Mining software engineering team project work logs to generate formative assessment

Supporting students in software engineering team project can be a challenge. Team leaders and supervisors may at times find it difficult to monitor students' progress and contribution. Students may either be falling behind or not contributing properly to the team. This paper describes a proposed solution that generate formative assessment feedback automatically using text data mining techniques. Various text similarity and machine learning techniques were explored and experimented to identify a suitable model for assessing student's performance and generating feedback. Utilising the students' individual weekly logs and the team's project plan, the proposed solution experimented on different text similarity techniques to match work done against work planned. The calculated similarity score and other extracted features are then applied to different machine learning algorithms, with the root mean-squared error used as the evaluation metric to identify the most suitable model. With this proposed solution, formative feedback generated can be used by team leaders and supervisors to identify team problems early on, and provide the students with necessary support. The students themselves can also reflect on their performance and address them earlier in the project phase than later.