Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper tackles the SympTEMIST NER and EL tasks by fine-tuning a RoBERTa-based token classifier with BiLSTM and CRF layers for NER, and employing cross-lingual SapBERT XLMR-Large for candidate generation in EL. Data augmentation is used to improve NER performance. The study finds that the selection of the knowledge base significantly impacts the accuracy of the entity linking component.
Knowledge base choice, not model architecture, is the surprisingly dominant factor in entity linking accuracy for the SympTEMIST dataset.
This paper presents a transformer-based approach to solving the SympTEMIST named entity recognition (NER) and entity linking (EL) tasks. For NER, we fine-tune a RoBERTa-based (1) token-level classifier with BiLSTM and CRF layers on an augmented train set. Entity linking is performed by generating candidates using the cross-lingual SapBERT XLMR-Large (2), and calculating cosine similarity against a knowledge base. The choice of knowledge base proves to have the highest impact on model accuracy.