Relation Classification via Recurrent Neural Network

Deep learning has gained much success in sentence-level relation classification. For example, convolutional neural networks (CNN) have delivered state-of-the-art performance without much effort on feature engineering as the conventional pattern-based methods. A key issue that has not been well addressed by the existing research is the lack of capability to learn %in modeling temporal features, especially long-distance dependency between nominal pairs. In this paper, we propose a novel framework based on recurrent neural networks (RNN) to tackle the problem, and present several modifications to enhance the model, including a max-pooling approach and a bi-directional architecture. Our experiment on the SemEval-2010 Task-8 dataset shows that the RNN model can deliver state-of-the-art performance on relation classification, and it is particularly capable of learning long-distance relation patterns. This makes it suitable for real-world applications where complicated expressions are often involved.