Most attempts to test spreadsheets use the spreadsheet-internal model to automatically detect input, intermediate and output cells.
In this paper, we discuss an example which shows why this approach is problematic even for simple spreadsheets. Following this, we derive a number of requirements for more feasible spreadsheet cell-meaning models and describe a first prototype we have designed.
Even a simple spreadsheet can have complex relationships between input, intermediate, and output cells. In this example, determining which cells are considered to be the outputs depends on the user and what they're using the spreadsheet for.
The proposed cell-meaning model, which separates input and output cells, permits users to specify cell meanings, and allows for different users to have different views of cell meaning.
2014, First Workshop on Software Engineering Methods in Spreadsheets, July