Semantics-aware BERT for Language Understanding

TL;DR

This paper introduces Semantics-aware BERT (SemBERT), which enhances language understanding by integrating explicit semantic information from semantic role labeling into the BERT model, leading to improved performance on multiple NLP tasks.

Contribution

The paper proposes a novel Semantics-aware BERT that incorporates structured semantic information, significantly improving upon standard BERT without complex modifications.

Findings

Achieves state-of-the-art results on ten NLP tasks.

Substantially improves performance over BERT.

Maintains ease of use with light fine-tuning.

Abstract

The latest work on language representations carefully integrates contextualized features into language model training, which enables a series of success especially in various machine reading comprehension and natural language inference tasks. However, the existing language representation models including ELMo, GPT and BERT only exploit plain context-sensitive features such as character or word embeddings. They rarely consider incorporating structured semantic information which can provide rich semantics for language representation. To promote natural language understanding, we propose to incorporate explicit contextual semantics from pre-trained semantic role labeling, and introduce an improved language representation model, Semantics-aware BERT (SemBERT), which is capable of explicitly absorbing contextual semantics over a BERT backbone. SemBERT keeps the convenient usability of its BERT precursor in a light fine-tuning way without substantial task-specific modifications. Compared with BERT, semantics-aware BERT is as simple in concept but more powerful. It obtains new state-of-the-art or substantially improves results on ten reading comprehension and language inference tasks.