Exploring collaborative writing of user stories with multimodal learning analytics: A case study on a software engineering course

Software engineering is the application of principles used in engineering design, development, testing, deployment, and management of software systems. One of the software engineering's approaches, highly used in new industries, is agile development. User stories are a commonly used notation to capture user requirements in agile development. Nevertheless, for the elaboration of user stories, a high level of collaboration with the client is necessary. This professional skill is rarely measured or evaluated in educational contexts. The present work approaches collaboration in software engineering students through multimodal learning analytics, modeling, and evaluating students' collaboration while they are writing user stories. For that, we used multidirectional microphones in order to derive social network analysis metrics related to collaboration (permanence and prompting) together with human-annotated information (quality of the stories and productivity). Results show that groups with a lower productivity in writing user stories and less professional experience in managing software requirements present a non-collaborative behavior more frequently, and that teams with a fewer number of interventions are more likely to produce a greater number of user stories. Moreover, although low experience subjects produced more user stories, a greater productivity of the most experienced subjects was not statistically verified. We believe that these types of initiatives will allow the measurement and early development of such skills in university students.