The research in i* should be with support of industry
Unless that we have some feedback about the advantages and disadvantages of the language, how can we think of the future of the research in i*.
Tools
Modeling concepts
We have an RE analysis, we can have on one side secure RE, and on the other side the quality issues. Can i* support all of these issues?
Add extra concepts to cover everything, what primitives are missing
Only add extensions which cover certain situations
There are many researchers working on different aspects of the usage of the language, e.g. the security, trust, and other functional and non-functional requirements. This issue deals with the coupling of i* with other areas of research, i.e. how closely relate the research in security (or performance,...) with research in i* modeling.
Is i* adequate for which domain?
I* doesn’t cover all the bases, can’t be used for everything
It is a very simplistic view to the language; the language should be used WITH some other modeling languages. We should not expect that this modeling language solve all of the problems of a domain (e.g. service oriented modeling, social modeling, etc.)
Instead of expansion, the solution is a combination of techniques
Analysis
Methodology
Metamodel and syntax is not enough, how do you use it?
KB (collection of models in a knowledge base)
Could be part of some sort of professional portal, which may have a charge for use
Coupling with other modeling methods, BPMN
Link to models covering reliability, security: research question of how to link to each of these
Foundations, philosophical underpinning
Model is a mean not an end.
Community – interactions and community building
We should do something to send i* out of our community.
Put i* literature paper on citeulike, <istar> tag
Hard to search for papers
Attracting new people
Possibility of working with MBA students for free on a project. They look for ways to demonstrate their marketing skills. How would you market i*?
We want to spread i* out of the community, we are experts, but need to help others use it
For RE courses, this could help, long-term investment
Students use different variants of i*, which is hard to control
Are you sure that this problem is not the same as with UML for every of its constructs.
Even for UML, almost no papers use the standard metamodel
Metamodel is useful for this, hard to teach students when there are so many versions “I found it this way in this paper!”
Need funding, but not feasible for standard research funding
Have some sort of group, where users pay some not barrier fee to join and gain access to information, money motivates action of consolidation
Unified language/metamodel:
Maybe we are in the 90’s in comparison with OO modeling and UML, maybe not ready?
Research will follow their own way either way
From adoption point of view, good to have a metamodel as a starting point
Could we have an online discussion about the metamodel propositions
Having an online discussion. We can post some of the major discussions (e.g. an vs the i* metamodel). We can use the i* wiki.
Take into account early adopters from industry? Early UML was motivated by some industrial partners
Since their modeling languages have strong industry support, they can advocate such approaches, but we cannot, we need to sit down and talk w beer
Such approaches can help to distribute i*
Need to find semantic definitions for things, not just core concepts. If language has redundancies, there will different approaches for the same thing and confusion.
Comments on the guidelines
Open version displayed on the same page to avoid clicking
People need to motivated to change the way they do things, with UML they were business people, made lots of money, we are more of a social collective.
Invoke different extensions as service, but things have to be standardized
It’s the added extras that make things useful. If we all had that power to access services directly, that would be helpful, incentive to adapt use of i*.
Unification; it is interesting to know that i* is a socialist community in comparison with UML which became commercialized. I would love to invoke, for example, Jennifer’s analysis as a web service, and then I can use that web service. If these web services become available then it will help to distribute the modeling language.
Process of developing i* and When should we use i*? The situations in which i* can be helpful, and those that it cannot be.
We are at a different level, can’t translate directly to code, so harder to standardize
Expansions to new areas
we are moving to social computing and i* can play an important role here
we should think of some proposals about different concepts that are being treated in i*
Ok, but we should be careful not to make the language more complicated, or it’s difficult to use in practice
to reconsider the i* we should consider the aspects of social computing
Presence of parallel research in social computing opens opportunities for grad students, finding case studies
Our problem with i* is that it is not executable, like UML, so we cannot make money out of it. Currently there is not a strong agent oriented programming language that the industry can be interested in.
You cannot make i* executable because it will be so detailed.
There are many different extensions to i*, it’s confusing to know what to pick in what situations, emphasis on the why for extensions
Most of the research on i* is based on phd thesis, thus you need to extend the language.
Closing thoughts
Maybe a special issue of something for paper expansions
I* book due on January 2011
Next workshop? Maybe with CAiSE next June in London
Opportunity for industry session in London as per 4 years ago
