Recurrent Entity Networks with Delayed Memory Update for Targeted Aspect-based Sentiment Analysis
This addresses fine-grained sentiment extraction for specific aspects, which is an incremental advance in natural language processing.
The paper tackles targeted aspect-based sentiment analysis by proposing a novel architecture with external memory chains and delayed memory updates, achieving substantial improvements over state-of-the-art approaches.
While neural networks have been shown to achieve impressive results for sentence-level sentiment analysis, targeted aspect-based sentiment analysis (TABSA) --- extraction of fine-grained opinion polarity w.r.t. a pre-defined set of aspects --- remains a difficult task. Motivated by recent advances in memory-augmented models for machine reading, we propose a novel architecture, utilising external "memory chains" with a delayed memory update mechanism to track entities. On a TABSA task, the proposed model demonstrates substantial improvements over state-of-the-art approaches, including those using external knowledge bases.