L'analisi testuale della disclosure finanziaria: dal machine learning al deep learning

This paper provides a scoping review of the literature on textual analysis in accounting — that is, the application of natural language processing to textual data to measure disclosure tone and readability, to determine similarities or differences in disclosure venues, to assess forward-looking statements, and to detect topics — with a focus on developments over the last decade. In the review, we analyze key contributions on the models developed by prior literature to analyze textual data in accounting, which are based on machine learning, and recently on deep learning. Finally, we provide an overview of areas in which our understanding is still limited and discuss opportunities for future research.