Semiconductor technology serves as a vital national economic asset and the foundation for advanced industries, making its trend analysis crucial for R&D strategies and technological opportunity identification. Although numerous semiconductor technology trend analyses have been conducted, systematic analysis of technology trends by linking technological problems and solutions included in patent documents to understand the context and direction of technology development remains insufficient. Therefore, this study identifies technological problems (Problem) and corresponding implementation approaches (Solution) from patent documents and proposes a Problem-Solution-based method for analyzing technology trends.

Specifically, we (1) collect domestic registered patents in the semiconductor field using a patent search formula, (2) extract Problem and Solution elements from the patent specifications using prompt engineering with large language models, (3) organize Problem and Solution area by clustering similar elements, and (4) construct a Problem-Solution network and positioning map to analyze semiconductor technology trends. By applying large language models and a Problem-Solution approach, this study presents a novel method for conducting technology trend analyses in the semiconductor field. This study analyzes problem-solving patterns by linking technical problems and solutions that comprise semiconductor technology. The results can contribute to supporting strategic decision-making by researchers and companies in the semiconductor field.