Text Sentiment Classification Algorithm Based on BiLSTM Deep Learning

In recent times, the swift evolution of mobile internet technology and the ubiquitous presence of diverse applications have made e-commerce and social media platforms integral to everyday life. These platforms offer a wide variety of products and services, facilitating user interactions through comment sections where consumers share their experiences. Typically, these comments are textual and rich in emotional expressions and personal perspectives. Accurate analysis of these emotional tendencies is crucial for platforms seeking to comprehend user requirements, refine service quality, elevate satisfaction levels, and execute targeted marketing strategies. To address this need, our study introduces a sentiment classification algorithm leveraging bidirectional long short-term memory (BiLSTM) and deep learning (DL) techniques. This algorithm captures the bidirectional context of text sequences using BiLSTM, enhancing the precision in identifying emotional cues. Experimental outcomes demonstrate that our proposed method outperforms traditional sentiment classification approaches, exhibiting notable gains in both accuracy and recall across various datasets.