Core Concepts
OATS dataset introduces new domains and comprehensive annotations for ABSA tasks, aiming to bridge gaps in existing datasets and enhance ABSA research.
Abstract
OATS dataset introduction and motivation
Aspect-Based Sentiment Analysis (ABSA) overview
Challenges in ABSA datasets and the need for OATS
Detailed analysis of OATS dataset structure and statistics
Performance evaluation of baseline methods on ABSA tasks using OATS
Significance of OATS dataset for ABSA research and future directions
Stats
OATS dataset includes 27,470 sentence-level quadruples and 17,092 review-level tuples.
ASQP dataset has 5,791 total sentences, with 1,578 positive, 210 negative, and 85 neutral sentiments.
BMRC method outperformed other baselines in ASTE task across all domains.
Quotes
"OATS dataset aims to bridge gaps in existing ABSA datasets and enhance research in the field."
"BMRC method emerges as the standout performer in the ASTE task, surpassing other baselines."