Daniati, Erna (57194168856); Wibawa, Aji Prasetya (56012410400); Irianto, Wahyu Sakti Gunawan (57208223701); Nafalski, Andrew (35587492800)
Event extraction from story narratives remains a challenging yet underexplored area in natural language processing due to narrative complexity including implicit causality long-range dependencies and temporal ambiguity. This study addresses the research question: How have NLP and deep learning approaches been applied to extract events from story narratives and what gaps persist. Following the PRISMA 2020 guidelines we systematically reviewed 12 peer-reviewed studies published between 2017 and 2024. Our analysis reveals growing adoption of transformer-based models such as BERT alongside emerging architectures like DEEIA and PAIE which leverage prompt-based learning and event-specific contextual aggregation. Commonly used datasets include ROCStories and custom narrative corpora though few are standardized. Key challenges involve handling implicit events limited annotated data cross-domain generalization and integration of commonsense reasoning. The main contribution of this review is the first structured synthesis of event extraction techniques specifically for story narratives using a rigorous systematic methodology. We highlight the need for document-level modeling narrative-aware evaluation metrics and low-resource adaptation strategies. This work provides a foundation for future research aiming to bridge narrative understanding with robust event-centric NLP systems. © 2026, Universitas Ahmad Dahlan. All rights reserved.
Department of Electrical Engineering and Informatics, Universitas Negeri Malang, Indonesia; Department of Informatics Engineering, University of South Australia, Australia
Research at a Glance
Register to unlockTopics & SDG Alignment
Register to unlockCollaboration
Register to unlockAuthor Profile (Selected)
Register to unlockReferences Overview
Register to unlockJournal & Source
Register to unlockMetadata & Integrity
Register to unlock