Question-Driven Span Labeling Model for Aspect–Opinion Pair Extraction

Aspect term extraction and opinion word are two fundamental subtasks of aspect-based sentiment analysis. The internal relationship between aspect terms words is typically ignored, information for the decision-making buyers sellers insufficient. In this paper, we explore an aspect–opinion pair (AOPE) task propose a Question-Driven Span Labeling (QDSL) model to extract all pairs from user-generated reviews. Specifically, divide AOPE into (ATE) aspect-specified (ASOE) subtasks; first candidate then corresponding given term. Unlike existing approaches that use BIO-based tagging scheme extraction, QDSL adopts span-based builds question–answer-based machine-reading comprehension effective extraction. Extensive experiments conducted on three tasks (ATE, ASOE, AOPE) four benchmark datasets demonstrate proposed method significantly outperforms state-of-the-art approaches.