We introduce an attention-based Bi-LSTM for Chinese implicit discourserelations and demonstrate that modeling argument pairs as a joint sequence canoutperform word order-agnostic approaches. Our model benefits from a partialsampling scheme and is conceptually simple, yet achieves state-of-the-artperformance on the Chinese Discourse Treebank. We also visualize its attentionactivity to illustrate the model's ability to selectively focus on the relevantparts of an input sequence.
展开▼