Tabular abstraction, editing, and formatting

This dissertation investigates the composition of high-quality tables with the use of electronic tools. A generic model is designed to support the different stages of tabular composition, including the editing of logical structure, the specification of layout structure, and the formatting of concrete tables. The model separates table's logical structure from its layout structure, which consists of tabular topology and typographic style. The notion of an abstract table, which describes the logical relationships among tabular items, is formally defined and a set of logical operations is proposed to manipulate tables based on these logical relationships. An abstract table can be visualized through a layout structure specified by a set of topological rules, which determine the relative placement of tabular items in two dimensions, and a set of style rules, which determine the final appearance of different items. The absolute placement of a concrete table can be automatically generated by applying a layout specification to an abstract line. An NP-complete problem arises in the formatting process that uses automatic line breaking and determines the physical dimension of a table to satisfy user-specified size constraints. An algorithm has been designed to solve the formatting problem in polynomial time for typical tables. Based on the tabular model, a prototype tabular composition system has been implemented in a UNIX, X Windows environment. This prototype provides an interactive interface to edit the logical structure, the topology and the styles of tables. It allows us to manipulate tables based on the logical relationships tabular items, regardless of where the items are placed in the layout structure, and capable of presenting a table in different topologies and styles so that we can select a high-quality layout structure.