A collaborative filtering based persona identification in requirements elicitation

Abstract

Persona is a fictional character that archetypically represents a user group. Persona identification is an important step in requirements elicitation. A review of related literature has shown that the persona is identified using qualitative approaches such as ethnographic profiling, user observations and user interviews. These approaches classify users on the basis of demographics or behavioral patterns. The drawbacks for such qualitative approaches are: they focus on detailed information gathering rather than correctly identifying representative user of persona; identified personas are too subjective as different requirements analysts may create different personas; these approaches do not scale well for a large number user involvement due to the high computational complexity of processing unstructured data. This paper proposed the collaborative filtering based persona-scenario (CFPS) approach to identify persona by calculating the similarities between the representative user to other users, combining the collaborative filtering algorithm and the persona-scenario approach. The case study shows the proposed approach improves the efficiency and accuracy in persona identification and requirements elicitation.