Consumer QoE-Aware Cognitive Semantic Sentiment Analysis via Hybrid Large Models
Haiyu Xu, Zhiwei Guo, Aldosary Saad, Amr Tolba, Anwer Al‐Dulaimi, Keping Yu, Joel J. P. C. Rodrigues
Abstract
Nowadays, the importance of Quality of Experience (QoE) has gained increasing attention from online consumers. For service providers, developing effective sentiment analysis approaches is essential to accurately capture and understand the emotional characteristics of consumers, thereby enhancing consumer QoE. However, existing research often encounters limitations in comprehending QoE and lacks fine-grained semantic analysis. To overcome these challenges, this article introduces a novel pretraining neural network structure designed for QoE-aware cognitive semantic sentiment analysis using hybrid large models. This approach utilizes a unique parallel architecture encoder that effectively captures implicit semantics within user reviews, significantly improving the model's ability to comprehend semantic sentiments. By learning diverse features, the model adeptly captures intricate word-level relationships, which enhances its generalization capabilities. Finally, the model's efficacy in sentiment analysis has been validated using real-world online consumer datasets in multilingual contexts, demonstrating its practical utility for online consumer sentiment analysis.