Incorporating Dynamic Semantics into Pre-Trained Language Model for Aspect-based Sentiment Analysis

03/30/2022
by   Kai Zhang, et al.
3

Aspect-based sentiment analysis (ABSA) predicts sentiment polarity towards a specific aspect in the given sentence. While pre-trained language models such as BERT have achieved great success, incorporating dynamic semantic changes into ABSA remains challenging. To this end, in this paper, we propose to address this problem by Dynamic Re-weighting BERT (DR-BERT), a novel method designed to learn dynamic aspect-oriented semantics for ABSA. Specifically, we first take the Stack-BERT layers as a primary encoder to grasp the overall semantic of the sentence and then fine-tune it by incorporating a lightweight Dynamic Re-weighting Adapter (DRA). Note that the DRA can pay close attention to a small region of the sentences at each step and re-weigh the vitally important words for better aspect-aware sentiment understanding. Finally, experimental results on three benchmark datasets demonstrate the effectiveness and the rationality of our proposed model and provide good interpretable insights for future semantic modeling.

READ FULL TEXT

page 1

page 2

page 3

page 4

research
03/22/2019

Utilizing BERT for Aspect-Based Sentiment Analysis via Constructing Auxiliary Sentence

Aspect-based sentiment analysis (ABSA), which aims to identify fine-grai...
research
12/02/2020

Exploiting BERT to improve aspect-based sentiment analysis performance on Persian language

Aspect-based sentiment analysis (ABSA) is a more detailed task in sentim...
research
10/31/2020

Understanding Pre-trained BERT for Aspect-based Sentiment Analysis

This paper analyzes the pre-trained hidden representations learned from ...
research
07/17/2022

Aspect-specific Context Modeling for Aspect-based Sentiment Analysis

Aspect-based sentiment analysis (ABSA) aims at predicting sentiment pola...
research
06/01/2023

Uncertainty-Aware Unlikelihood Learning Improves Generative Aspect Sentiment Quad Prediction

Recently, aspect sentiment quad prediction has received widespread atten...
research
08/06/2021

LadRa-Net: Locally-Aware Dynamic Re-read Attention Net for Sentence Semantic Matching

Sentence semantic matching requires an agent to determine the semantic r...
research
03/22/2022

BERT-ASC: Auxiliary-Sentence Construction for Implicit Aspect Learning in Sentiment Analysis

Aspect-based sentiment analysis (ABSA) task aims to associate a piece of...

Please sign up or login with your details

Forgot password? Click here to reset