A Unified Tagging Solution: Bidirectional LSTM Recurrent Neural Network with Word Embedding
This provides a general-purpose tagging method for NLP researchers and practitioners, reducing the need for task-specific feature engineering, though it is incremental as it builds on existing BLSTM-RNN techniques.
The paper tackled the problem of multiple tagging tasks (part-of-speech tagging, chunking, and named entity recognition) by proposing a unified solution using a Bidirectional LSTM-RNN with word embeddings, achieving nearly state-of-the-art performance across all tasks.
Bidirectional Long Short-Term Memory Recurrent Neural Network (BLSTM-RNN) has been shown to be very effective for modeling and predicting sequential data, e.g. speech utterances or handwritten documents. In this study, we propose to use BLSTM-RNN for a unified tagging solution that can be applied to various tagging tasks including part-of-speech tagging, chunking and named entity recognition. Instead of exploiting specific features carefully optimized for each task, our solution only uses one set of task-independent features and internal representations learnt from unlabeled text for all tasks.Requiring no task specific knowledge or sophisticated feature engineering, our approach gets nearly state-of-the-art performance in all these three tagging tasks.