Development of an intelligent system for automatic response to customer reviews on marketplaces with using LLM

The article discusses an approach to the development of an intelligent system designed for automatic responses to customer reviews on marketplaces, utilizing advanced artificial intelligence technologies, specifically large language models (LLMs). The relevance of this work is driven by the growing need for automating customer interactions for e-commerce companies and reducing the time spent processing a large volume of reviews, which requires the application of high-tech solutions. The problem faced by companies is the need to respond quickly and effectively to a large number of reviews, which in a traditional setup demands significant effort from support staff. The methods include data collection and preprocessing, as well as the use of the GPT-4 large language model based on transformer architecture for sentiment analysis and generating personalized responses. The model analyzes the emotional tone of reviews and formulates responses that consider context, tone, and subtext, making communication with customers more natural and effective. The research results demonstrated a significant improvement in customer interaction quality, a reduction in review processing time, and a decrease in operator workload. The conclusions confirm that implementing this system in a company can significantly enhance the quality of customer service.