Visual Structure Editing of Math Formulas
Gavin Nishizawa · RIT Scholar Works (Rochester Institute of Technology) · 2020
Math formulas can be large and complex resulting in correspondingly large and complex LaTeX math strings for expressing them. We design operations to visually edit the typeset LaTeX formulas. The operations are invoked via the formula's control points, which are created as a way to specify an operation associated with the point's location relative to a symbol in the formula. At the control points, formulas can be extended in multiple ways, LaTeX can be inserted locally by typing, an existing formula can be inserted, or part of the formula itself can be moved to that point. Parts of formulas can be selected by clicking on a symbol or dragging a rectangle over an area in the formula, and the subtree for the selection can be replaced, deleted, moved to another point in the formula, or lifted out of the formula into a chip floating above the canvas. Formula chips can be used as arguments to operations, including a set of existing formulas provided in a symbol palette. Operations can be performed either by making a selection, selecting a control point operation, and then specifying an argument, or by dragging an argument to one of the control points in the formula. We perform an online formula editing experiment to examine if these visual editing operations can be used to reduce the time and actions spent in order to make edits to formulas. With 35 participants completing 18 formula editing tasks split between 3 input conditions of LaTeX only, Visual only, or LaTeX and Visual, we find that on average participants spend the least amount of time on the editing tasks when both editing capabilities are available.