GNN and q-ROFN Fuzzy MCDM for Sentiment-Driven Product Comparison and Selection in Multi-Vendor e-Commerce Platforms

In the fast-growing e-commerce sector, product comparison and recommendation systems play an important role in improving user experience. Conventional models are unable to effectively incorporate sentiment analysis, especially for knowing customer preference and product attributes. This paper introduces a Graph Neural Network (GNN) with Q-Rung Ortho Pair Fuzzy Multi-Criteria Decision-Making (MCDM) for sentiment-based product comparison and selection on multi-vendor e-commerce websites. The new GNN model works outstandingly to classify products as positive, neutral, or negative sentiment classes with 98% accuracy, 95.32% precision, 96.67% recall, and 94.67% F1-score. Comparing with current models like BiRNN-LSTM, BERT, SLCABG, SVM, and DC-BiLSTM-CNN, the GNN model beats all of them in all measures, suggesting it as a good choice for product ranking and recommendation. Combination of sentiment analysis based on GNNs and fuzzy MCDM improves decision-making by balancing both quantitative and qualitative aspects of comparison, and it provides an active and customer-focused method to compare e-commerce products.