[Adapt] Does syntax matter? A strong baseline for Aspect-based Sentiment Analysis with RoBERTa

[朱煜烨]

Hi Adapters,

This week, I introduce you with a NAACL 2021 paper  "Does syntax matter? A strong baseline for Aspect-based Sentiment Analysis with RoBERTa". 

Aspect-Based Sentiment Analysis (ABSA) is a fine-grained task in the field of sentiment analysis, which aims at predicting the polarities for aspects. Generally, ABSA contains Aspect Extraction (AE) and Aspect-Level Sentiment Classification (ALSC). In this paper, both dependency parsing tree and Pre-Trained Models (PTMs) are studied. The authors also compare and analyze the induced trees from PTMs and the dependency trees for the task. 

The code can be found here: https://github.com/ROGERDJQ/RoBERTaABSA

Time: Wed 4:00pm

Venue: SEIEE 3-414

Best Regards,

Yuye Zhu