This study is the first-year component of a three-year project aimed at establishing an internationally comparable social security fiscal analysis framework based on OECD SOCX. Using Large Language Models (LLMs), we constructed an integrated database that combines expenditure levels (quantitative data) and institutional characteristics (qualitative data) for 6,637 programs across 41 countries. A pilot analysis of the old-age domain demonstrated that this methodology enables the generation of integrated fiscal indicators, assessment of Korea's position through international comparison, and application to long-term fiscal analysis. This study is significant in establishing a fiscal analysis infrastructure capable of overcoming the limitations of expenditure-centered approaches and enabling simultaneous analysis of institutional characteristics and expenditure levels.