DomainTermsViz: A Microsoft Excel add-on to reduce the error-proneness of spreadsheets

In the course of my PhD research work at the University of Eastern Finland School of Computing, I also developed a Microsoft Excel add-on, dubbed DomainTermsViz, whose purpose is to help spreadsheet creators to map easily application domain concepts to their spreadsheet-specific counterparts. The Microsoft Excel add-on translates traditional spreadsheet formulae into application domain concepts while also automatically highlighting all cells referenced in an active translated spreadsheet formula. All formula cells in a spreadsheet are also automatically marked with magenta right border markings. Just as any Microsoft Excel add-on, this add-on will alter the behaviour of normal Microsoft Excel. However, it is important that this change is done in a non-invasive way so that the environment should not be completely different from the traditional spreadsheet environment. This is to take advantage of existing user experience as familiarity with existing tools and techniques could become worthless if the add-on is too different from what spreadsheet creators know and how they are used to work. Moreover, spreadsheet users are known to be very conservative when it comes to adopting new features.

An illustration of the add-on in action: Formula in cell B9, which is =B4 +B5,  has automatically been translated to "Cash + Inventories" (Note also that cells B4 and B5 are also missing column names, otherwise the column names would have been included in the translation).

The add-on has the potential to be used in the following spreadsheet activities:

1. When creating or editing a spreadsheet, the add-on can help one in tracking effected changes as active formula cell ranges are highlighted and a corresponding problem domain description is displayed for an active formula cell. This has potential in helping to create an error free spreadsheet as it will be providing an intuitive on-spot feedback to the spreadsheet developer on changes being done to the spreadsheet. This feedback can thus act as a verification aid hence introducing a non-invasive way of introducing some kind of spreadsheet testing to spreadsheet creators, who being non professional programmers are not interested in formal verification of their spreadsheets. As the adage says "prevention is better than cure", it is important to prevent errors from occurring in spreadsheets.
2. The add-on may also be helpful in the comprehension of a spreadsheet created by others through visual comprehension aids such as highlighting all formulae in a spreadsheet and for each formula cell, a click on it will highlight its precedent cells (formula range) and a corresponding problem domain based description of the cell is given in the formula cell caption box.
3. The add-on may also help in the comprehension of a spreadsheet as it can give a user a general overview of the structure of a spreadsheet through the magenta right border markings on all formula cells.
4. The add-on may also be helpful in locating errors in a spreadsheet. For example, a problem domain description or narrative for a given formula cell, that does not match with what is expected in the problem domain could provide a visual cue for an error in that formula cell.

The Microsoft Excel add-on has been tested to run in Microsoft Excel 2003, Microsoft Excel 2007 and Microsoft Excel 2010. It can be downloaded freely at http://cs.uef.fi/spreadsheet_tools/ together with installation instructions. Note: Download the first add-on on the list! You may try the add-on and your feedback is most welcome.  

Read more...

A summary of my PhD thesis in not so technical language

My PhD thesis defence was held on 28th August 2015 at 12 noon at the University of Eastern Finland School of Computing in the Louhela Auditorium, Joensuu Science Park in Joensuu, Finland. The opponent was Research Fellow Sari Kujala and the custos was Professor Jorma Sajaniemi. A press release of my doctoral thesis to the Finnish media (as required by Finnish academic tradition) is at https://www.uef.fi/en/-/28-8-virheita-taulukkolaskennassa-onko-aika-muuttaa-taulukkopohjien-tekotapoja- 

The title of my PhD thesis is "Deficiencies in Spreadsheets: A Mental Model Perspective". The thesis can be downloaded at http://epublications.uef.fi/pub/urn_isbn_978-952-61-1828-4/index_en.html  Below is a summary of the thesis in not so technical language: 

Millions of people, globally, use spreadsheets in their daily work. For example, information extracted from spreadsheets is used in decision making and planning in various businesses and organizations. Spreadsheet software is ubiquitous such that almost all desktop computers have some spreadsheet software installed on them. Their easy to use interface, coupled with their powerful data manipulation capabilities, has made spreadsheets a darling to many end-users. Despite this popularity of spreadsheets, different researchers have shown that most spreadsheets have errors. Unfortunately, these errors have proven to be not trivial. For example, it is documented how some companies have lost millions of dollars due to errors in the spreadsheets they use. Research, however, has also shown that most spreadsheet errors involve construction and use of formulae.

This thesis tackled this problem of errors by studying mental models of spreadsheet creators performing various spreadsheet activities. In other words, we opted to study what spreadsheet creators think when they are doing various spreadsheet activities. We opted for this approach because it is a common assertion that humans have mental models of the systems they interact with, and as a result, it is difficult to explain many aspects of human behaviour when interacting with systems, without resorting to a construct such as mental models.

Using a constructive research approach, we conducted three empirical studies and also developed two Microsoft Excel add-ons. Spreadsheet creators are normally not professional programmers but are experts in their application domains such as accounting. However, the results of the studies showed that with traditional spreadsheets, spreadsheet creators have to link up concepts from their application domain to low-level spreadsheet concepts such as cell references in order for them explain a spreadsheet as well as to locate and fix errors in spreadsheets. Therefore, we created a Microsoft Excel add-on that translates traditional spreadsheet formulae into application domain concepts so that spreadsheet creators should map easily application domain concepts to their spreadsheet-specific counterparts. The add-on also automatically highlights all cells which are referenced in a translated spreadsheet formula. The add-on can be used for different purposes such as being used as a verification aid when creating a spreadsheet, as a comprehension aid when trying to understand a spreadsheet created by others, and as a debugging aid for locating potential errors in a spreadsheet.

The add-on was evaluated and found to be easy to learn and also helped spreadsheet creators to locate more errors in their spreadsheets. The add-on was also noted to help spreadsheet creators to think more in terms of the application domain rather than in terms of spreadsheet specific concepts. This is important as spreadsheet creators are experts in their application domains and are not necessarily professional programmers and as such they are more familiar with application domain concepts rather than low-level spreadsheet specific concepts.

We also developed another Microsoft Excel add-on (a prototype one) by extending the previously described Microsoft Excel add-on such that, instead of using cell references as in traditional spreadsheets, spreadsheet creators could now debug their spreadsheets using application domain concepts only. An evaluation of this add-on showed that it correspondingly promoted spreadsheet creators to think more in terms of their application domain such that they almost no longer thought of their spreadsheet in terms of spreadsheet-specific low-level concepts. This has the implication that we can have a spreadsheet paradigm in which spreadsheet creators develop their spreadsheets using application domain concepts only rather than labouring themselves to learn to use traditional cell references. In other words, spreadsheet creators would develop their spreadsheets in the usual ad-hoc, incremental manner but instead of using traditional cell references, they would use application domain concepts, of which they are already familiar of.

The two add-ons have been tested to run in Microsoft Excel 2003, Microsoft Excel 2007 and Microsoft Excel 2010. They can be downloaded freely at http://cs.uef.fi/spreadsheet_tools/ together with installation instructions.

Read more...