Does this persona represent me?: investigating an approach for automatic generation of personas based on questionnaires and clustering

Abstract

Personas are fictional representations of an end user based on data collected from a specific target audience. The creation of personas takes time, because part of the process is manual, which causes difficulties during the stages of data generation and analysis, if the data sample is large. The present work aims to investigate the results of the automatic generation of personas through a questionnaire and the Clustering method. Initially, a questionnaire was applied in order to outline the profile of students in the Computer Science and Software Engineering courses at the university Federal of Ceará. 130 responses were obtained from this applied questionnaire. The automatic generation of personas used the collected responses as a database, these data were applied to the Clustering method. From Clustering, using a specific tool for this purpose, it was possible to generate four personas automatically. In relation to the data generated from the personas, it was possible to identify some characteristics, such as: (a) course choice factor: the opportunity to work with new technologies, influence of family and friends and the possibility of living abroad; (b) factor of withdrawal from the course: lack of identification with the course, learning difficulties in the disciplines and finance; and, (c) area of interest: software engineering, human-computer interaction and computer graphics. A new questionnaire was applied in order to validate the personas generated in relation to credibility, empathy and similarity. Based on the results obtained, it was observed that two of the four personas generated achieved greater prominence in relation to the validation criteria, showing that the personas look like real people, students think like the personas and share similar interests.