A Hierarchical Structure Explanation Method for Complex Tables

Fangnuo Liu, Sainan Tong, Derong Shen, Tiezheng Nie, Yue Kou

Abstract: Tables are used to organize and manage data. It is important to extract valuable information from tabular data. Existing methods for interpreting table structure focus on relational tables, but have limitations. In our paper, we propose a method for interpreting complex tables hierarchically. We use cell functional classification(CFCT) and a tree-based region detection model (RDT) to classify cells accurately. We also introduce a table layout prediction model(LPT) based on relational reasoning to explore the spatial structure of tables. Experiments on real datasets show the effectiveness of our approach.