Multilingual Named Entity Recognition Using Pretrained Embeddings, Attention Mechanism and NCRF

Abstract: In this paper we tackle multilingual named entity recognition task. We use the BERT LLM as embeddings with bidirectional recurrent network, attention, and NCRF on the top. We apply multilingual BERT only as embedder without any fine-tuning. We test out model on the dataset of the BSNLP shared task, which consists of texts in Bulgarian, Czech, Polish and Russian languages.