<![CDATA[Event Extraction (EE) is a pivotal task for NLP, where important events in the narrative text need to be detected and recognized. We present an alternative method for extracting events from Hans Christian Andersen's fairy tales, utilizing Few-Shot Learning with BERT (Bidirectional Encoder Representations from Transformers) and RNN (Recurrent Neural Network) in this paper. We selected Andersen's fairy tales because they are characterized by rich narratives and symbolic language, which also often prevents automatic event extraction. To reduce reliance on labeled samples, we utilize the Few-Shot Learning method, which enables the model to learn from a small number of labeled event examples trivially. The BERT model is used to generate deep representations by modeling the context between words and sentences. RNN is essential to capture the sequence of events in the story, which determines the structure of the narrative. The findings demonstrate that the proposed framework significantly improves event extraction, with high values of evaluation metrics such as in accuracy, precision, recall, and F1-score. The proposed method is also effective in extracting non-explicit events while keeping the narrative context. Despite the challenges posed by metaphorical language and subjective events, this work demonstrates that Few-Shot Learning, BERT, and RNNs offer a promising solution to the task of event extraction from complex narratives.]]>
Copyrights © 2025
