Aspect-based Sentiment Analysis for Vietnamese

Multi-task Solution for Aspect Category Sentiment Analysis (ACSA) on Vietnamese Datasets

• Demo: https://youtu.be/ggmAvpA4oHQ

Overview

We solved 2 tasks in the Vietnamese Aspect-based Sentiment Analysis problem: Aspect Category Detection (ACD) and Sentiment Polarity Classification (SPC). Besides, we proposed end-to-end models to handle the above tasks simultaneously for 2 domains (Restaurant and Hotel) in the VLSP 2018 ABSA dataset using PhoBERT as Pre-trained language models for Vietnamese in 2 ways:

• Multi-task (ACSA-v1):
• Multi-task with Multi-branch approach (ACSA-v2):

The models achieved good results when concatenating the last 4 layers of BERT together. You can download the model weights here.

Dataset

• The VLSP 2018 Aspect-based Sentiment Analysis dataset:

Domain	Dataset	Reviews	Aspects	AvgLength	VocabSize	DiffVocab
	Training	2,961	9,034	54	5,168	-
Restaurant	Dev	1,290	3,408	50	3,398	1,702
	Test	500	2,419	163	3,375	1,729
	Training	3,000	13,948	47	3,908	-
Hotel	Dev	2,000	7,111	23	2,745	1,059
	Test	600	2,584	30	1,631	346

• Preprocessing:

flowchart LR
A[Remove\nHTML] --> B[Standardize\nUnicode] --> C[Normalize\nAcronym] --> D[Word\nSegmentation] --> E[Remove\nunnecessary\ncharacters]

Loading

Results

Task	Method	Hotel			Restaurant
		Precision	Recall	F1-score	Precision	Recall	F1-score
Aspect Detection	VLSP best submission	76.00	66.00	70.00	79.00	76.00	77.00
	Bi-LSTM+CNN	84.03	72.52	77.85	82.02	77.51	79.70
	BERT-based Hierarchical	-	-	82.06	-	-	84.23
	Multi-task	87.45	78.17	82.55	81.09	85.61	83.29
	Multi-task Multi-branch	63.21	57.86	60.42	80.81	87.39	83.97
Aspect + Polarity	VLSP best submission	66.00	57.00	61.00	62.00	60.00	61.00
	Bi-LSTM+CNN	76.53	66.04	70.90	66.66	63.00	64.78
	BERT-based Hierarchical	-	-	74.69	-	-	71.30
	Multi-task	81.90	73.22	77.32	69.66	73.54	71.55
	Multi-task Multi-branch	57.55	52.67	55.00	68.69	74.29	71.38