Prompt Chain Engineering for Customer Satisfaction Analysis via Large Language Models

In the contemporary era of technological evolution, the swift progress of large language models (LLMs) has democratized the realm of data analysis, rendering it more accessible to a broader audience. This work delves into the innovative application of LLMs within the sphere of satisfaction analysis, with a primary objective of empowering non-specialists to harness the potential of these advanced models for their analytical endeavors. The core of this study lies the concept of prompt chain and prompt engineering, which serves as a strategic approach to guide the interaction between users and LLM. By meticulously delineating the operational procedures, parameter configurations, and nuanced use cases of LLMs in satisfaction analysis, this paper establishes a comprehensive prompt framework. This framework is designed to be user-friendly, enabling non-experts to navigate the complexities of LLMs with relative ease. To further validate the efficacy of this approach, we integrate SPSS statistical analysis as a benchmark for comparison. The performance and outcomes of satisfaction analysis conducted via the designed prompt chain are rigorously compared with those obtained through traditional SPSS methods.