Improving Span Representation by Efficient Span-Level Attention

High-quality span representations are crucial to natural language processing tasks involving span prediction and classification. Most existing methods derive a span representation by aggregation of token representations within the span. In contrast, we aim to improve span representations by considering span-span interactions as well as more comprehensive spantoken interactions. Specifically, we introduce layers of span-level attention on top of a normal token-level transformer encoder. Given that attention between all span pairs results in O(n 4 ) complexity (n being the sentence length) and not all span interactions are intuitively meaningful, we restrict the range of spans that a given span could attend to, thereby reducing overall complexity to O(n 3 ). We conduct experiments on various span-related tasks and show superior performance of our model surpassing baseline models. Our code is publicly available at https://github.com/jipy0222/ Span-Level-Attention.