Complex Industrial Tables for Structure-Aware Large Language Modeling

This dataset includes 12 tables collected from real-world applications, which not only represent a wide range of application fields but also exhibit structural diversity, ranging from relatively standardized formats to highly complex layouts. These tables are used to evaluate the performance of understanding the contents of tables by using a large language model with retrieval augmentation generation (RAG).  These tables are categorized into five major domains.: 1) Engineering Domain, 2) Manufacturing Industry, 3) Corporate Human Resource and Internal Management, 4) Social Statistics and Policy Analysis, and 5) Corporate Sustainability.