LLM과 ABSA를 활용한 주거환경만족도 영향요인 분석: 노후계획도시 정비 선도지구를 중심으로 = Analyzing Factors Influencing Residential Environment Satisfaction Using Large Language Models and Aspect-Based Sentiment Analysis: Focusing on Leading Redevelopment Districts in Aging Planned Cities

다국어 초록 (Multilingual Abstract)

With the designation of leading redevelopment districts in aging planned cities in Korea, understanding residents’ perceived residential experiences has become increasingly important. This study aims to identify the key factors influencing residential environment satisfaction in apartment complexes located in these districts. Resident review data from the online real estate platform Zigbang were analyzed using a Large Language Model (LLM)-based Zero-shot Aspect-Based Sentiment Analysis (ABSA). The extracted aspects and sentiments were further interpreted in conjunction with GIS spatial data. The results indicate that residential environment satisfaction can be categorized into three dimensions: accessibility, management, and residential environment. Accessibility to public transportation and parks showed positive perceptions, while parking shortages and inter-floor noise were identified as major sources of dissatisfaction. This study proposes a mixed-method analytical framework integrating LLM-based ABSA with GIS spatial analysis to better understand residents’ perceptions of residential environments in aging planned cities.