Improving Constituency Parsing with Span Attention

TL;DR

This paper introduces span attention with n-gram weighting for neural constituency parsing, significantly improving performance across multiple languages by better capturing contextual information, especially for long spans.

Contribution

It proposes span attention with n-gram weighting and categorical span attention to enhance span representations in neural chart-based constituency parsing.

Findings

Achieved state-of-the-art results on Arabic, Chinese, and English datasets.

Effectively models long-span information with categorical span attention.

Demonstrates significant performance improvements over existing methods.

Abstract

Constituency parsing is a fundamental and important task for natural language understanding, where a good representation of contextual information can help this task. N-grams, which is a conventional type of feature for contextual information, have been demonstrated to be useful in many tasks, and thus could also be beneficial for constituency parsing if they are appropriately modeled. In this paper, we propose span attention for neural chart-based constituency parsing to leverage n-gram information. Considering that current chart-based parsers with Transformer-based encoder represent spans by subtraction of the hidden states at the span boundaries, which may cause information loss especially for long spans, we incorporate n-grams into span representations by weighting them according to their contributions to the parsing process. Moreover, we propose categorical span attention to further enhance the model by weighting n-grams within different length categories, and thus benefit long-sentence parsing. Experimental results on three widely used benchmark datasets demonstrate the effectiveness of our approach in parsing Arabic, Chinese, and English, where state-of-the-art performance is obtained by our approach on all of them.