Context-Based Deep Learning Approach for Named Entity Recognition in Hindi

Named Entity Recognition is a crucial step in Natural Language. For a language like Hindi, this task is difficult due to complex linguistic features. To overcome the difficulties posed by the nature of the language we propose a deep learning method for performing Named Entity Recognition in Hindi. In our method, we use contextualized word embeddings and a Bi-LSTM neural network. The results observed have an average accuracy of 98.67%. The model successfully captures the context of ambiguous words and classifies them correctly.