Assessment of Service Quality of Online Fresh Food Service Platforms: Leveraging Large Language Model Information Extraction Technology and Trustworthy Knowledge Aggregation Algorithms

Abstract With the rapid development of e‐commerce in the fresh agricultural products sector, while it has brought convenience to consumers, it has also led to quality‐related issues such as spoilage and adulteration. Online complaint platforms, such as Black Complaint, offer consumers a public and real‐time channel to report problems. This study aims to develop a systematic online service quality assessment model for e‐commerce of fresh foods. To achieve this objective, the paper proposes an innovative multistep approach by integrating large language models (LLMs) technology with information theory. Specifically, the first step involves using LLM prompting techniques and the Chain of Thought method to classify the attributes of the complaint text and assess the levels of consumer satisfaction from user‐generated complaint texts. This enables precise classification and multidimensional analysis of complaints. Subsequently, expert evaluation criteria—covering dimensions such as delivery timeliness, product quality, and category richness—are incorporated to ensure a comprehensive service quality assessment. To effectively integrate both expert evaluations and consumer feedback, the basic uncertain interval q‐Rung orthopair fuzzy information and its aggregation operator are employed. A service quality attribute‐based attitude characteristic target optimization model is then constructed to address uncertainties in expert assessments and consumer opinions. Finally, through a comparative analysis of four main fresh e‐commerce platforms, the proposed model demonstrates superior performance in practical applications. This research not only overcomes the limitations of existing service quality evaluation methods but also provides a systematic and actionable framework to improve the service quality of fresh e‐commerce platforms.