As technology continues to advance, one area that has gained substantial momentum is the application of artificial intelligence (AI) and automated customer service in the context of new retail. Automated customer service systems, powered by AI, are capable of engaging with customers through chatbots, virtual assistants, or voice recognition technologies. These systems not only provide immediate assistance and support, but they also serve as a valuable tool for capturing customer sentiments and opinions during their interaction with the system. By understanding customer sentiments, retailers can identify areas of strength and weakness in their products, services, and overall shopping experience. This article presents a comment sentiment analysis model (CSA) and an automated customer service system for gathering feedback from new retail consumers. This methodology can improve the new retail business by inferring the audience's opinion towards it through emotional analysis of remarks. In order to retrieve the most fundamental characteristics of the comment text, the model first extracts attribute words from the comments, then filters and clusters the collected candidate attribute words. Second, LSTM is limited in its ability to model long-term dependencies and leverage bidirectional sequence information due to its inability to parallelize training. Lightweight multi-head self-attention mechanism and gated convolutional neural network are the basis for the model proposed in this paper. Finally, the CSA model's viability was tested by a series of experiments presented in this paper.