Unleashing the True Potential of Sequence-to-Sequence Models for Sequence Tagging and Structure Parsing

TL;DR

This paper systematically explores sequence-to-sequence models for structure prediction tasks, demonstrating that constrained decoding with diverse linearization schemas can outperform models relying on external resources, achieving state-of-the-art results.

Contribution

The study introduces novel constrained decoding methods with diverse linearization schemas that enable S2S models to effectively predict complex structures without extra parameters.

Findings

Constrained decoding improves S2S performance on core tasks.

Lexical diversity in schemas affects training complexity and learning ease.

Best models match or surpass state-of-the-art results.

Abstract

Sequence-to-Sequence (S2S) models have achieved remarkable success on various text generation tasks. However, learning complex structures with S2S models remains challenging as external neural modules and additional lexicons are often supplemented to predict non-textual outputs. We present a systematic study of S2S modeling using contained decoding on four core tasks: part-of-speech tagging, named entity recognition, constituency and dependency parsing, to develop efficient exploitation methods costing zero extra parameters. In particular, 3 lexically diverse linearization schemas and corresponding constrained decoding methods are designed and evaluated. Experiments show that although more lexicalized schemas yield longer output sequences that require heavier training, their sequences being closer to natural language makes them easier to learn. Moreover, S2S models using our constrained decoding outperform other S2S approaches using external resources. Our best models perform better than or comparably to the state-of-the-art for all 4 tasks, lighting a promise for S2S models to generate non-sequential structures.