Aspect-Based Sentiment Analysis of FABSA (Feedback Aspect-Based Sentiment Analysis) Dataset using BERT Embeddings and Bidirectional Long Short-Term Memory

Aspect-Based Sentiment Analysis (ABSA) offers fine-grained insights into user opinions by linking sentiments to specific aspects within text. However, existing models often struggle to capture both contextual meaning and sequential dependencies of information within a sentence. This becomes a problem when working with real-world data, where user reviews can vary significantly in terms of complexity and structure. This study proposes a hybrid deep learning model that combines BERT embeddings with Bidirectional Long Short-Term Memory (BiLSTM) to improve contextual and sequential understanding in ABSA. Experiments on the FABSA dataset, comprising 10,574 annotated reviews across 10 domains and 12 aspect categories, show that the BERT-BiLSTM model outperforms the baseline BERT model, particularly in terms of recall and F1-score. The model achieves an average Aspect Category Detection f1-score of 0.75 and an average Aspect Category Sentiment Classification accuracy of 0.93, demonstrating the effectiveness of integrating sequential modeling into transformer-based architectures. These findings have practical implications for improving automated customer feedback systems, enabling more accurate sentiment tracking across diverse domains, and potentially enhancing sentiment analysis tasks.