Abstract

Following the COVID-19 pandemic, the gap between the consumer price index and perceived inflation has widened. Despite the significant increase in demand for timely information and the need to compare short-term indices with the consumer price index, an opportune index for online shopping prices is currently absent. To meet these growing demands, this study conducts foundational research to develop a timely price index that reflects the expansion of the online market and evolving consumer behaviors by integrating big data processing and artificial intelligence (AI) technology. To overcome the temporal and spatial limitations of current market methodologies, a novel approach utilizing large language models (LLMs) is proposed for processing and clustering unstructured data. The study also incorporates meta-learning methodologies with generative AI models to enhance the training data construction process and model maintenance. Utilizing over 71 million records from four major online sites, this research proposes a methodology for automating the process of structuring data. it also explores the possibility of subdivision, and identifies trends in short-term price variations by analyzing average price changes in subdivisions. Additionally, it provides a framework for further progressive research to develop a timely price index that can be compared with the consumer price index.