Aspect-level sentiment analysis of e-commerce reviews is important for identifying fine-grained customer experience signals that are often hidden by overall star ratings. This study proposes a dual-track framework for USA-based Amazon customer reviews, combining text-driven multi-aspect sentiment classification with structured-feature satisfaction prediction. For the text-based task, reviews are modeled across three operational aspects: Product & Value, Delivery & Fulfillment, and Service & Usability, each using a three-class sentiment scheme of Negative, Neutral, and Positive. The proposed deep learning model integrates DeBERTa-v3 contextual embeddings, bidirectional recurrent refinement, and aspect-specific additive attention to separate overlapping sentiment cues within the same review. Three recurrent variants, Bi-GRU, Bi-RNN, and Bi-LSTM, are evaluated under the same modeling framework. On the held-out Amazon Customer Reviews test set, the DeBERTa-v3 + Bi-GRU + aspect-attention model achieves the best overall performance, with 0.737 accuracy, 0.757 macro-F1, and 0.760 weighted-F1, outperforming the Bi-LSTM and Bi-RNN variants. Ablation analysis further shows that sequential refinement and aspect-specific attention improve performance over encoder-only and shared-attention configurations. In the structured-feature satisfaction task, LightGBM achieves the strongest result among the evaluated machine learning models, with 82.37% accuracy and an F1-score of 0.82, outperforming XGBoost and CatBoost. The results show that the proposed framework provides complementary insights: structured models support overall satisfaction prediction, while the DeBERTa-v3-based aspect model provides more detailed attribution of customer experience across product, delivery, and usability dimensions. Overall, the study offers a reproducible and interpretable approach for fine-grained review analytics in e-commerce settings.
Modern digital infrastructure generates heterogeneous, multimodal software artifacts encompassing structured code metrics, unstructured textual data such as commit logs and inline comments, and dynamic runtime signals whose complexity renders traditional defect detection approaches increasingly inadequate. This paper presents a cloud-native, transformer-based multimodal intelligence framework that integrates Large Language Model (LLM) semantic encoding with deep neural learning to enable automated defect prediction and proactive maintenance of large-scale digital infrastructure. The proposed system employs a sentence-transformer encoder (all-MiniLM-L6-v2) to process multimodal software artifact data—including serialized structured metrics and available textual fields—into dense 384-dimensional semantic embeddings. These embeddings are subsequently refined through a hierarchical multi-layer perceptron (MLP) deployed on a scalable cloud architecture for real-time inference. Evaluated on a real-world dataset of approximately 60,000 software modules, the framework achieves 99.72% accuracy, 100% precision, and an ROC-AUC of 0.9998, substantially outperforming baseline models including Random Forest, XGBoost, LSTM, and standalone MLP architectures. To address potential concerns regarding result validity, we conducted repeated experiments with five different random seeds (42, 0, 1, 7, 123) and ten-fold stratified cross-validation, confirming that performance metrics are stable across runs (accuracy: 0.9972 ± 0.0003). Data leakage was ruled out through strict temporal split ordering and pre-split SMOTE application exclusively on training folds. Confusion matrix and threshold-based analyses confirm strong classification performance with minimal false positives. SHAP-based explainability analysis further enhances the trustworthiness of the system by identifying the most influential multimodal predictors—past defect history, static analysis signals, and cyclomatic complexity—thereby contributing to transparent and accountable AI-driven infrastructure management. The presented framework advances the state of the art in LLM-driven multimodal systems by demonstrating how transformer intelligence, when applied to heterogeneous software artifact data streams, can enable reproducible, cloud-scalable, and interpretable maintenance pipelines for complex digital environments.
Mst Masuma Akter Semi, Md Masud Karim Rabbi, K. Ahmed et al.· Multimedia· 0 citations