This paper focuses on the computational identification of characters in fictional narratives, regardless of their nature, i.e., either humans, animals or other type of beings. We approach this problem as a supervised binary classification task, whether or not a noun in a narrative —specifically in a fairy tale— is classified as a character. A wide range of Machine Learning algorithms and configurations were tested in order to come up with the most appropriate model (or setof models) to successfully fulfil this task. Despite the challenges associated with the character identification in the domain of children stories, the best models obtain an F-Measure above 0.80, proving a good performance and broadly out performing the baselines.
Revista: 2019 IEEE International Conference on Cognitive Computing (ICCC)
Autores: Cristina Barros, Marta Vicente and Elena Lloret