Character-Based Named Entity Recognition in Keras

Using a Bi-Directional LSTM Recurrent Neural Network

Keras implementation based on models by:

• Kuru, Onur, Ozan Arkan Can, and Deniz Yuret. CharNER: Character-Level Named Entity Recognition.

• Klein, D., Smarr, J., Nguyen, H., & Manning, C. D. (2003, May). Named entity recognition with character-level models. In Proceedings of the seventh conference on Natural language learning at HLT-NAACL 2003-Volume 4 (pp. 180-183). Association for Computational Linguistics.

Usage

• Implement get_texts(), get_labels() and get_x_y() or get_x_y_generator() with your own data source.

  • x is a tensor of shape: (batch_size, max_length). Entries in dimension 1 are alphabet indices, index 0 is the padding symbol.

  • y is a tensor of shape: (batch_size, max_length, number_of_labels). Entries in dimension 2 are label indices, index 0 is the null label.

• Tweak the model hyper-parameters in config.py

• Run train.py