Learn from Structural Scope: Improving Aspect-Level Sentiment Analysis with Hybrid Graph Convolutional Networks
This work addresses the problem of noisy opinion words in aspect-level sentiment analysis for natural language processing applications, representing an incremental improvement through hybrid syntactic modeling.
The paper tackled aspect-level sentiment analysis by introducing the concept of Structural Scope to model relations between targets and sentiments based on grammatical hierarchy, and proposed a hybrid graph convolutional network (HGCN) that outperformed state-of-the-art baselines on four public datasets.
Aspect-level sentiment analysis aims to determine the sentiment polarity towards a specific target in a sentence. The main challenge of this task is to effectively model the relation between targets and sentiments so as to filter out noisy opinion words from irrelevant targets. Most recent efforts capture relations through target-sentiment pairs or opinion spans from a word-level or phrase-level perspective. Based on the observation that targets and sentiments essentially establish relations following the grammatical hierarchy of phrase-clause-sentence structure, it is hopeful to exploit comprehensive syntactic information for better guiding the learning process. Therefore, we introduce the concept of Scope, which outlines a structural text region related to a specific target. To jointly learn structural Scope and predict the sentiment polarity, we propose a hybrid graph convolutional network (HGCN) to synthesize information from constituency tree and dependency tree, exploring the potential of linking two syntax parsing methods to enrich the representation. Experimental results on four public datasets illustrate that our HGCN model outperforms current state-of-the-art baselines.