By David Stefan, Mark Barrett, Emmanuel Letier (University College London) and Mark Stella-Sawicki (Logica)
Abstract- Large organizations play an important role in helping to mitigate and adapt to the consequences of climate change. As a result, they face increasing pressure from Governments and non-governmental organizations to report on the sustainability of their operations. Beyond simple reporting, however, it is difficult for them to identify the most effective actions to take to address the risks associated with climate change and rising energy costs. The problem is hard because it involves tradeoffs between multiple long- term and short-term objectives that must be made under strong budgetary constraints, uncertainties about the future evolution of many system variables, and sometimes simply a lack of shared understanding of what the real objectives are and the potential impacts of various decisions on such objectives. The overall aim of our research is to develop fundamental techniques to help organizations make deci- sions in this context. As a first step, we are currently applying quantitative goal-oriented requirements engineer- ing technique to model and reason about the sustainability goals of UCL, a large university in central London. The paper also discusses important software systems engineering research challenges in this domain. These are related to the elaboration and evolution of large-scale models, the ability to reason about the timing of system transformations and the delayed impacts of these transformations on goals, and the need to include in the system model funding mechanisms and system governance structures as explicit components that are themselves subject to changes.
5 comments
Comments feed for this article
May 19, 2011 at 5:42 pm
Elena Pérez-Miñana
an interesting research proposal worth discussing at the workshop, specially given its application of “quantitative” goal-oriented requirements engineering. One of the most important challenges associated to this study is the management of the uncertainty associated to the data that has to be manipulated, there was no mention to this specific aspect, and I think it is a point worth discussing as it certainly needs to be properly tackled if the aim is to achieve the deployment of a quantitative decision support system. Another important element that was not mentioned is the plans to support the continuous “change of heart” regarding environmental policies that has become “de facto” behaviour of many governments.
June 7, 2011 at 12:46 pm
Steffen Zschaler
A very interesting paper indeed. Goal-oriented techniques, seem really very useful in this context and the paper points out a number of relevant research directions. Apart from the issue of uncertainty of data pointed out by Elena, I feel that it would have been useful to provide some more detail as to what you were already able to learn about UCL’s goals and strategies from the exercise as it stands. It would be interesting to compare this to what was learned in the approach that used i*. Clearly one would expect to be able to express goals and dependencies more precisely, but this also seems to imply more effort in maintaining and evolving the model. So the question arises how this added expressiveness helps in making decisions about and achieving/maintaining environmental goals.
September 7, 2011 at 7:43 pm
Martin Mahaux
I know I’m a bit late here, but I LOVE this idea of transparent review system, and I think it’s even better if the discussion can continue outside the conference, doesn’t it ?
First, I agree it’s a very interesting paper, and a very interesting way to follow.
Then, as author of what is cited as one of the “i* approaches”, I’d like to comment a bit on this.
The first thing is, we have not used i* at all. We used a lightweight KAOS paradigm. We even used “Objectiver”, a tool that draws KAOS models. It is possible that in some way it is not “100% proper” KAOS, but definitely it’s not i*. So, this is just wrong.
The authors then say that qualitative analysis offers “a very limited support” for decision making. Obviously, this is (at least) a biased view, or said with the wrong words. I’d rather say it offers a *different* support than quantitative one. Doing so, instead of claiming to be in some way “superior” to what we did, authors would claim complementarity, which is much closer to reality, and more constructive.
Whether you want to put the effort on quantitative modelling or not depends on what you need to achieve. In our case we mostly wanted clarify what “Be sustainable” meant for our client. We also wanted to discover more sustainability requirements, and be able to reason qualitatively about their priority. The quantitative analysis would have been not only unaffordable, but also useless in our case…
September 8, 2011 at 8:21 am
Steffen Zschaler
Thanks Martin for these comments. Could you perhaps elaborate a little more why quantitative analysis was not useful in your case? I do get the argument that it may have been unaffordable, but supposing you had had infinite money, would quantitative analysis not potentially have provided better results?
Just curious,
Steffen
September 23, 2011 at 8:50 pm
Martin Mahaux
Late again, sorry I did not register to the RSS feed…
As I said, we used the goal tree to 1) produce new ideas, and 2) clarify the meaning of sustainability for our client, and then finally to prioritize and make decisions. I realize now that this might sound a bit unorthodox, but it works quite well… Indeed, goal models are not only there for decision making, they also help in ensuring completeness as well as fostering communication and even creativity (I should write a paper about the latter, maybe it would be controversial enough
).
For 1) we had a generic goal refinement for the goal “sustainability”, and we tried to instantiate any of the leafs of this goal tree to our concrete case. So, to name a stupid example, if one leaf of the generic tree is “reduce waste”,we tried to see how we could do this in our context, and thus simply refined this generic goal within our project context.
For 2), well refining together the main “sustainability” goal with our client turned out to be a great way to have him talk about it extensively.
Then yes, prioritization has always to happen, so here we might have been tempted by quantitative analysis. Our feeling was that the qualitative work was then enough to say that some sub-goal was more important than another, because it was related to a more important goal. When you have a tree with 30 nodes or so, I guess it remains manageable. Or maybe was it just by chance. Or maybe I would have been surprised by a quantitative analysis result… So yeah, let’s admit I’m not SURE quantitative analysis would have been useless if I had many money I would not know what to do with it…
Cheers,
Martin.