Learning Explicit and Implicit Structures for Targeted Sentiment\n Analysis

2019/09/17 by Hao Li, Wei Lu, Li, Hao +2 · 2 citations
Computer Science · #Sentiment Analysis and Opinion Mining #Text and Document Classification Technologies #Topic Modeling

paper · pdf · doi:10.48550/arxiv.1909.07593

Abstract

Targeted sentiment analysis is the task of jointly predicting target entities\nand their associated sentiment information. Existing research efforts mostly\nregard this joint task as a sequence labeling problem, building models that can\ncapture explicit structures in the output space. However, the importance of\ncapturing implicit global structural information that resides in the input\nspace is largely unexplored. In this work, we argue that both types of\ninformation (implicit and explicit structural information) are crucial for\nbuilding a successful targeted sentiment analysis model. Our experimental\nresults show that properly capturing both information is able to lead to better\nperformance than competitive existing approaches. We also conduct extensive\nexperiments to investigate our model's effectiveness and robustness.\n

Cited by

• Position-Aware Tagging for Aspect Sentiment Triplet Extraction
• Exploiting BERT for End-to-End Aspect-based Sentiment Analysis

Related

• FCKT: Fine-Grained Cross-Task Knowledge Transfer with Semantic Contrastive Learning for Targeted Sentiment Analysis
• Learning in Order! A Sequential Strategy to Learn Invariant Features for Multimodal Sentiment Analysis
• Interactive Attention Networks for Aspect-Level Sentiment Classification
• Targeted Sentiment Analysis: A Data-Driven Categorization
• Multi-Task Learning with LLMs for Implicit Sentiment Analysis: Data-level and Task-level Automatic Weight Learning