A Conflict Degree-Guided Quantum Group Consensus Method Based on Graph Convolutional Networks With Enhanced Sentiment Analysis for Consumer Electronics Products

Chao Zhang, Xinyu Liang, Arun Kumar Sangaiah, Deyu Li

Abstract: Artificial intelligence-based human-computer interaction (AI-HCI) and consumer electronics (CE) are closely linked, making AI-HCI crucial in e-commerce by providing intelligent and personalized product recommendations. In this context, consumers often select from multiple websites and products when purchasing CE products. This makes CE product selections conflicting and random to reach a consensus, often modeled as a quantum group consensus problem. Therefore, this paper aims to enhance consumer selections via intelligent decision support using online reviews. First, an ensemble technique, multi-objective salp swarm optimization bi-directional long short-term memory (MOSSO-BiLSTM) with gate recurrent unit (GRU), is employed for sentiment analysis (SA). Second, graph convolutional networks (GCN) is used to calculate weights for key indicators. Meanwhile, according to hybrid distances, an improved fuzzy C-means three-way (FCMTW) clustering analysis is explored. Third, a conflict degree-guided quantum two-stage consensus reaching process (CRP) is designed. It includes identification rules determined by quantum frameworks and modification rules determined by regret theory (RT). Fourth, multi-objective optimization by ratio analysis plus the full MULTIplicative form (MULTIMOORA) and RT are combined for stable optimal product selections. Finally, a case study of selecting solar emergency lights verifies the viability of the proposed methodology.

External IDs:dblp:journals/tce/ZhangLSL25