Automatic table layout and formatting

In this thesis we address the problem of tabular layout with a focus on automatic table layout optimization. In a world where documents are produced in both physical format and also delivered electronically to an increasing range of smart devices, we draw attention to the fact that authoring tables is still di cult, frustrating, prone to errors and automated optimal table layout is basic or non-existent. We show through examples that while tables are well supported by document authoring tools, table layout su ers from many usability bottlenecks and inconsistencies. The current tools provide powerful features for controlling the presentation parameters but they provide limited or no features for controlling the logical table structure. In this context we survey table creation in several popular document authoring programs and review previous work on logical table models. We then present a combinatorial optimization modelling method for the table layout optimization problem, the problem of minimizing a table's height subject to it tting on a given page (width). We present three models of the problem for tables with spanning cells and inner tables and evaluate their performance using mixed integer programming, constraint programming and local search. When a table is too large to be displayed in one page, a multiple-objective optimization problem arises for automated table layout, nding suitable ways of splitting the table across multiple pages without a ecting the meaning of the data. We present two mixed integer programming optimization models for nding arrangements of columns in pages such that the the number of pages is minimized and, at the same time, the changes to the column or row ordering or grouping is also minimized. To demonstrate our results, we developed a Table Drawing Tool prototype which implements an automated solution for the table layout optimization problem using a mathematical modelling method.