基于多模态的新能源汽车用户需求挖掘

Multimodal-Based User Demand Mining for New Energy Vehicles

摘要: 本文针对新能源汽车市场快速扩张背景下，消费者在线评论中多模态信息的情感分析需求，提出了一种基于改进多头注意力机制的多模态情感分析模型。该模型通过跨模态和自注意力机制的融合，有效提升了新能源汽车在线评论中情感倾向的识别精度。实验结果表明，该模型在多个数据集上的性能优于现有方法，为新能源汽车用户需求挖掘提供了新的视角和工具。

Abstract: This paper addresses the need for sentiment analysis of multimodal consumer reviews in the rapidly expanding new energy vehicle market by proposing an improved multi-head attention mechanism-based multimodal sentiment analysis model. The model effectively enhances the recognition accuracy of emotional tendencies in online reviews by integrating cross-modality and self-attention mechanisms. Experimental results demonstrate that the model outperforms existing methods across multiple datasets, providing a new perspective and tool for mining user requirements in new energy vehicles.