Towards Resource-Constrained Event Extraction: A Knowledge-Augmented Framework for Overcoming Challenges in Vietnamese NLP

Abstract

Event Extraction (EE) is crucial for NLP, capturing meaningful activities in documents. While recent research has applied Large Language Models to EE, their computational overhead remains problematic. Moreover, existing state-of-the-art methods predominantly focus on high-resource languages such as English and Chinese, leaving low-resource languages, like Vietnamese, largely under-explored due to unique linguistic challenges. We propose a Small Language Model-based framework enhanced with external knowledge to address Vietnamese EE. The approach targets core difficulties including rare events, semantic ambiguity, and long-range dependencies, thereby establishing an efficient and robust framework specifically tailored for the Vietnamese low-resource language domain.