Real-Time Review Analysis with a Sentiment-Integrated Hybrid LDA Approach
Students & Supervisors
Student Authors
Supervisors
Abstract
In the era of rapid digital communication, real-time understanding of consumer sentiment is essential for businesses seeking to enhance their products and services. This study proposes a Sentiment-Integrated Hybrid Latent Dirichlet Allocation (LDA) model that improves traditional topic modeling by embedding emotional context directly into the topic generation process. Conventional LDA models often struggle with short, dynamic texts commonly found in social media and online reviews. To address this, the proposed approach incorporates sentiment filtering via a BERT-based classifier, which refines topic distributions and enhances thematic coherence. The model is evaluated on a diverse consumer review dataset, demonstrating improved performance in terms of topic coherence (0.497) and perplexity (-5.58) compared to standard LDA. The results indicate that the model effectively identifies sentiment-rich and contextually relevant themes, offering a deeper insight into consumer opinions. Due to its scalability and adaptability, the model is well-suited for applications such as social media monitoring, customer feedback analysis, and decision support systems. This work contributes to the advancement of sentiment-aware artificial intelligence and provides a practical framework for extracting actionable insights from large-scale textual data.
Keywords
Publication Details
- Type of Publication:
- Conference Name: 4th IEEE International Conference on Robotics, Automation, Artificial-Intelligence and Internet-of-Things (RAAICON-2025)
- Date of Conference: 27/11/2025 - 27/11/2025
- Venue: Military Institute of Science and Technology (MIST), Dhaka, Bangladesh
- Organizer: IEEE Bangladesh Section and IEEE Robotics and Automation Society Bangladesh Chapter