Exploring Opportunity of Large Language Model for Job Vacancy Extractions

In recent years, the growth of Online Job Vacancies (OJVs) has provided a valuable opportunity for researchers and practitioners to gain insights into the Job Market. However, this opportunity has not been effectively utilized in Indonesia, leading to issues such as the mismatch between curriculum and industry needs. This research aims to explore the use of Large Language Models (LLMs) to extract information from Online Job Vacancies, focusing on key variables such as skills and tools. By developing a tailored Prompt Chain, this research successfully extracted relevant information from 312 OJVs. The evaluation was conducted using semantic precision and recall metrics, achieving precision scores of 97.783% for skills and 99.140% for tools, and recall scores of 90.167% for skills and 95.071% for tools. This evaluation was validated by a LinkedIn Top Voice 2024, who assessed the semantic similarity between the extracted outputs and the established ground truth. This expert-guided evaluation highlights the potential of LLMs in Labor Market Intelligent tools and job matching systems. The findings contribute to understanding the strengths and limitations of LLMs in semantic text extraction, offering insights into their future scalability and adaptation across various job sectors.