Study on the economic value of BERT model in consumer sentiment analysis and market demand forecasting

Smart cities optimize urban management through big data and artificial intelligence, improve the efficiency of resource allocation, and promote the accuracy of market demand forecasting and consumer behavior analysis. As the impact of consumer sentiment on market demand changes becomes increasingly significant, traditional market demand forecasting methods face numerous challenges in responding to fluctuations in consumer sentiment. Therefore, this study combines BERT model for consumer sentiment analysis and integrates it with the KANO model to enhance the accuracy and efficiency of market demand forecasting. The research results show that the BERT-KANO model performs excellently in sentiment analysis and demand forecasting tasks. During the training process, the accuracy of BERT-KANO reaches 98%, and the loss rate drops to 0.005, far exceeding the performance of traditional models such as LSTM-GRU and CNN-RNN. Sentiment analysis results indicate that after-sales service and ease of operation receive high positive evaluations from consumers, with scores of 0.86 and 0.84 respectively, while brand reputation and service life show strong negative sentiment, with scores of -0.21 and -0.38 respectively. Improvements in these areas are crucial for enhancing market competitiveness. The sentiment analysis and demand forecasting method combining BERT model and KANO model can effectively improve the accuracy of market demand forecasting, especially in identifying consumer sentiment and demand intensity.