Authors
Yirsaw Ayalew & Ethel Tshukudu
Abstract
Spreadsheets, programs developed by end-user programmers, are used for a variety of important tasks and decisions. However, as the literature indicates, a significant proportion of spreadsheets contain faults.
One of the difficulties in understanding and debugging spreadsheets is the invisibility of data dependencies associated with cell formulas. To address this issue, we developed a graph based visualization tool based on the Markov Clustering (MCL) algorithm.
The prototype tool, which has been integrated into Microsoft Excel, provides a visualization of a spreadsheet in terms of its data dependency graph using a cluster tree. In addition, it highlights groups of cells that belong to a cluster with unique color and border style on the original spreadsheet. Using the visualization tool, spreadsheet users may narrow their focus to one cluster (i.e., logical unit) at a time.
This paper discusses the results of a controlled experiment conducted to investigate the effectiveness and efficiency of the prototype tool. We used cognitive fit theory as the basis for the evaluation of the tool. Among the features of the tool, highlighting of clusters was found to be useful for spreadsheet debugging while data dependency graph based visualization did not improve effectiveness and efficiency of debugging a spreadsheet.
Sample
We asked the subjects to what extent the tool helped them in the fault identification task. Subject's responses were mostly positive.
However, the results show that subjects who used the visualisation tool did not perform better than those without the tool. In other words, the tool did not improve the effectiveness of identifying faults and did not assist in minimizing the time needed to identify faults.
Publication
2016, International Journal of Software Engineering and Its Applications, Volume 10, Number 3, pages 107-118
