Researchers have proposed a new approach to parsing academic documents called Compilable Academic Document Parsing (CADP). This method reconstructs a full page of an academic paper as contextual LaTeX code plus executable Python code. The goal is to preserve the structure and data of elements like tables, formulas, and pseudocode, making it easier for machines to understand and verify scientific knowledge. A benchmark called CADP-Bench has been created to test this approach, which includes a set of expert-verified academic pages with tightly coupled text and multiple structured elements. The results show that even state-of-the-art models struggle to produce high-fidelity executable reconstructions, indicating room for improvement in structure-aware scientific document parsing. --- Why it matters: This matters because it could improve the accuracy and efficiency of scientific knowledge extraction from academic papers, which is crucial for applications like research collaboration, citation analysis, and automated summarization. It also highlights the need for more advanced models that can handle complex structured data. Source: https://arxiv.org/abs/2608.17550