PersonaBase: A Formative Evaluation Study of Computational Resources for Data-Driven Personas

Abstract

Drawing inspiration from the natural language processing community’s definitions of tasks, datasets, metrics, and baselines, we present PersonaBase, a repository of computational resources for data-driven persona (DDP) research. PersonaBase contains (1) computational notebooks that present algorithmic approaches for persona generation and evaluation, (2) persona generation prompts, (3) persona systems, and (4) datasets with user data or personas. Formative evaluation with four domain experts suggests that PersonaBase addresses a real need and suggests that DDP benchmarking can be further developed through community building and learning from other computing sciences.