Enhancing NEV Brand Equity Through Big Data Analytics: An LDA-LSTM Approach to Mining Online Consumer Reviews

Enhancing brand value is critical for new energy vehicle (NEV) enterprises amid fierce competition. This study leverages online consumer reviews as core big data to drive brand equity improvement via advanced big data analytics. A large-scale dataset of 5564 reviews for top five best-selling NEVs was collected from "Dongche Di" via web scraping, followed by a big data processing pipeline (data cleaning, Jieba segmentation, and stop-word filtering). To mine unstructured text big data, we used word cloud visualization, semantic network analysis, and an Latent Dirichlet Allocation (LDA)-Long Short-Term Memory (LSTM) fusion model: LDA identified key consumer concern dimensions, while LSTM enabled deep sentiment classification. Big data analysis revealed five core NEV brand perception dimensions (range, driving experience, interior space, price, and high-speed performance) and quantified emotions-prominent negativity in driving experience, minimal negativity in interior space, and overall dominant negativity. Guided by the Consumer-Based Brand Equity model, we proposed brand enhancement strategies. This study showcases big data analytics' power in scaling consumer perception understanding, offering a data-centric framework for NEV firms to optimize branding.