Portuguese Named Entity Recognition Using LSTM-CRF

摘要

Named Entity Recognition is a challenging Natural Language Processing task for a language as rich as Portuguese. For this task, a Deep Learning architecture based on bidirectional Long Short-Term Memory with Conditional Random Fields has shown state-of-the-art performance for English, Spanish, Dutch and German languages. In this work, we evaluate this architecture and perform the tuning of hyper-parameters for Portuguese corpora. The results achieve state-of-the-art performance using the optimal values for them, improving the results obtained for Portuguese language to up to 5 points in the Fl score.