Relation Extraction (RE) is a fundamental task in Natural Language Processing (NLP) crucial for constructing knowledge graphs and enhancing information retrieval. While significant progress has been made in monolingual and cross-lingual RE, the unique challenges posed by code-switched (mix-lingual) text remain largely underexplored due to a scarcity of dedicated datasets and tailored methodologies. This paper introduces a novel, large-scale dataset specifically designed for code-switched relation extraction. Furthermore, we propose an effective training methodology tailored to capture the complexities of inter- and intra-sentential code-switching phenomena. Our comprehensive experiments demonstrate that this new dataset and the proposed approach significantly advance the state-of-the-art in extracting relations from mix-lingual content, providing a valuable resource and benchmark for future research in this challenging domain.
📖 افتح في inklap 🔗 DOI 📮 اطلب بحثاً