Automatic Evaluation Metrics for Enhancing the Quality of Automatic Story Generation in NLP

Louisa Camadini

Abstract: In recent years, Automatic Story Generation (ASG) has become a popular subfield of Natural Language Processing (NLP). However, to ensure that ASG models produce high-quality outputs, reliable evaluation methods are necessary. Human evaluation, though effective, can be expensive and time-consuming. As a result, researchers have shifted their focus to developing Automatic Evaluation Metrics (AEM) that can accurately assess the quality of ASG outputs and also align with human judgement. This research area is essential for the progression of ASG and has become a prominent topic in NLP research.