Another great systems engineering conference in Paris with AFIS the French chapter of INCOSE.
Organised since 1998 by Jean-Claude Rault of Génie Logicel, and introduced this year by Alain Roussel, Chairman of AFIS INCOSE France, this international conference had delegates from many countries including UK, Germany, Sweden, Algeria, Austria, and of course France.
Presentations and discussions came from both universities and industry and coverd a wide range of areas of interest including large computers and also real-time embedded systems.
Highlights included a treatise by Christian Winkler on Requirements Engineering and Natural Language Processing. Not only looking at the left side of the V, Christian explored natural language in all environments. With good humor we enjoyed the history from hundreds of years of natural language processing leading to todays challenges. Can a machine deduce meaning from text with the accuracy and subtlety of a human? One of the geniuses that opened the field in the 1930’s was Alan Turing. See the Turing Test from 1950. It seems that in spite of massive progress there is still work to do. Look up WordNet, Lesk, and Noam Chomsky if you are further interested in this topic.
On the right side of the V Vassilk Kirova assessed the advancement of testing practices. There are a plethora of capability models for assessment of testing departments and test engineers. One of the most interesting parts of any technological discussion is how things work in practice. How are assessments received by those affected, and do they help to motivate people or do they act as a distractor? Nokia, building on experience gained over years from Lucent and Alcatel, are moving to a map of practices rather than the shorter and concise levels of testing capability. It is suggested that some people are tempted to concentrate on achieving the highest assessment of testing capability level rather than concentrating on achieving the correct level of verification for the given situation.
Colin Hood presented The V-Model is Dead, Long Live the V-Model. The discussion was rich and informative. Accepted opinions were challenged and some people understood that their reluctance to accept the V-model was due to self-imposed restrictions that are not part of the V-model. The V-model helps to understand what the system under development must do, and to check whether these aims are achieved. The V-model helps to divide large challenges into smaller challenges, including specifying a system of systems, and to build systems from smaller parts. As an example of agile, Scrum was projected onto the V-model, showing how the requirements for features in the backlog can be developed iteratively and incorporated incrementally into the system ready for deployment.