Browse Theses

Throughout the history of the spreadsheet, and throughout the majority of research into improving it, the grid of cells has remained a constant as the underlying data model. An idea that has received recent interest is to provide users with a spreadsheet-like environment based on something other than a grid. The attraction is that if salient features of the data structure can be made more explicit, the machine will be able to provide certain types of error checking and automation. In this project I consider one such grid replacement, a new data model which I call the "lish". It is based on nested lists of cells, composed according to rules that allow repeating structures to be described. It allows columns, tables, groups of tables and other structures to be treated as coherent objects. This supports a novel form of cell range selection, and allows the machine to ensure that related structures are kept consistent. The model is also more accommodating than the grid of dynamic space allocation, where the number of cells occupied by a result is not known in advance. Then, I develop a "lish calculus", an extension to vector arithmetic for hierarchical structures that provides a concise notation for calculations with lishes. This simplifies the usual spreadsheet formula expressions, and enables the machine to interpret them consistently with the context in which they are located. I evaluate the lish in the framework of the cognitive dimensions of notations, with the help of example use cases and a user study based on a prototype lish editor. These verify many of the hypothesised advantages, but also reveal some difficulties for users. I close with an analysis of how the lish might be revised to address these shortcomings, while continuing to capitalise on the essential benefits.