This issue presents three novel solutions to quite difficult problems. Stochastic modelling and simulation approaches to analysing enhanced fault tolerance on service-based software systems, by Kuan-Li Peng and Chin-Yu Huang, presents two approaches to improve the effectiveness of fault tolerant analysis. (Recommended by Min Xie.) Exhaustive test sets for algebraic specifications, by Marc Aiguier, Agnès Arnould, Pascale Le Gall, and Delphine Longuet, presents a way to generate test sets from formal specifications that are not only complete, but exhaustive. (Recommended by Lu Zhang.) Using combinatorial testing to build navigation graphs for dynamic web applications, Wenhua Wang, Sreedevi Sampath, Yu Lei, Raghu Kacker, Richard Kuhn, and James Lawrence, presents a new technique to address the very difficult problem of finding navigation structures within web applications. This is difficult because much of the navigation in web applications is not revealed unless the user enters specific input values. (Recommended by Jane Hayes.)
A common research dissemination practice is to expand a relatively short conference paper into a longer paper to submit to the journal. This practice is sometimes encouraged by special issue invitations and sometimes proactively started by authors. Regardless of how it starts, the authors must decide how to reference the original conference paper in the journal version. This editorial gives policy guidance for STVR.
The first question is “how much extension is enough?” That of course varies, but the usual rule of thumb is taken from common publisher guidelines: a new paper should have at least 30% significant new material over and above any material published elsewhere. The word “significant” is intended to apply to technical contribution, and the 30% is judged by the editors and reviewers.
While this 30% rule allows a 10-page conference paper to be significantly extended to a journal paper, it precludes the practice of combining several conference papers into one journal paper without adding additional new results.
Once determining that the longer paper has sufficient new material, the authors need to decide what to say about the previous paper. STVR expects the following policies to be followed:
This is consistent with the journal’s view that missing references are seldom grounds for rejecting a paper. The one exception is if a previous paper contains the same results, making the submitted paper a repetition. And an exception to that could be a replicated experiment, which of course can be quite valuable even if the results are very similar.
A model that STVR does not use is that of including information about the original paper in a letter to the editors as opposed to the paper itself. Pragmatically, that approach takes extra work and attention from the editor-in-chief, the reviewing editor, and the reviewers. That creates three opportunities for the information in the letter to get lost. Second, the information about the previous paper would not be present in the published paper. Third, the reference gives credit to the original conference (and the original paper). If the editors notice that a submitted paper is based on a previous paper, but not described as per these policies, we will send the paper back to the authors with a request to add this information before sending it out for review.
I want to acknowledge Lionel Briand, Gregg Rothermel, and Rob Hierons for discussions leading to this policy.
[1] Jeff Offutt. Who is An Author? (Editorial), Wiley’s journal of Software Testing, Verification, and Reliability, 25(2), March 2015.
Jeff Offutt
George Mason University
offutt@gmu.edu
13 April 2016