Title: Industrial Topics in Urban Labor System

Abstract: Categorization is an essential component for us to understand the world for ourselves and to communicate it collectively. It is therefore important to recognize that classification system are not necessarily static, especially for economic systems, and even more so in urban areas where most innovation takes place and is implemented. Out-of-date classification systems would potentially limit further understanding of the current economy because things constantly change. Here, we develop an occupation-based classification system for the US labor economy, called industrial topics, that satisfy adaptability and representability. By leveraging the distributions of occupations across the US urban areas, we identify industrial topics - clusters of occupations based on their co-existence pattern. Industrial topics indicate the mechanisms under the systematic allocation of different occupations. Considering the densely connected occupations as an industrial topic, our approach characterizes regional economies by their topical composition. Unlike the existing survey-based top-down approach, our method provides timely information about the underlying structure of the regional economy, which is critical for policymakers and business leaders, especially in our fast-changing economy.