The Monument of the Discoveries in Lisbon celebrates the Portuguese Age of Discovery during the 15th and 16th centuries, when Portuguese ships departed from the Tagus River to explore and trade with India and the Orient, thereby getting to know new worlds. With the same adventurous spirit, researchers and practitioners in the field of Requirements Engineering meet twice a year to discuss the current practical and research challenges in shaping the software systems of the future. The first meeting this year took place in Essen at REFSQ 2017 (23rd International Working Conference on Requirements Engineering: Foundation for Software Quality, 27 Feb-2 Mar) under the motto “My RE”. The motto emphasized the issue of how RE can evolve to deal with engineering diverse types of systems of different scale and complexity belonging to various domains. The second meeting took place in Lisbon at RE 2017 (25th International Requirements Engineering Conference, 4-8 Sep) and adopted the theme “Desperately Seeking Less: The Role of Simplicity and Complementarity in Requirements”. Now, with the Monument of Discoveries of the beautiful city of Lisbon still in mind, it is time to reflect on the directions given by these two meetings.
Further Predominance of Natural Language
Despite the diversity of graphic models available for performing RE and the relevance of model-based development, most requirements are still stated in some form of natural language, with varying degrees of structure. In his keynote at REFSQ, Lionel Briand explained why this is the case and reported on his experience performing collaborative research on Natural Language Processing with industrial partners to address several relevant challenges. Confirming the importance of dealing with natural language, REFSQ had two further sessions on Natural Language Processing: one on the extraction of requirements and another on automated quality checking.
Towards Crowd RE
Crowd RE stands for performing requirements engineering with the support of a crowd of stakeholders, usually current or potential users, in an automated way. The term was coined in 2015 (https://sse.uni-due.de/crowdre15/) when Fraunhofer IESE organized the first international workshop on Crowd RE, collocated with RE 2015. The vision consists of using two complementary mechanisms: User Feedback Analysis and Usage Mining. The first implies text analysis of feedback given by users of a target and/or competitor software. A classic application is the analysis of feedback given in app stores. The second implies monitoring the use of a target software to identify bottlenecks for its users. Both mechanisms aim at extracting requirements in order to evolve the target software. At RE 2017, Crowd RE appeared as a consolidated area of research in RE with two dedicated sessions and several other paper presentations scattered over the conference.
Tangled Requirements and Usability Engineering
This direction is represented by the fact that both meetings had keynotes on Design Thinking, where the understanding of the users and their viewpoints go together with prototyping and testing. Moreover, REFSQ 2017 featured a talk on “Bridging the gap: Requirements Engineering meets Usability Engineering”, which led to a discussion about the border between the two disciplines. The trend is a thin and malleable border, if any at all, with professionals who have a rich spectrum of expertise: at the edges, there are professionals with strengths in one of the disciplines but enough knowledge in the other one, and in the middle there are professionals with a balanced knowledge of both disciplines.
Security Requirements
The number of paper presentations at RE 2017 discussing specific approaches for dealing with security requirements (8 presentations) deserves attention. The presentations addressed the nature of security requirements, the derivation of privacy policies, security requirements engineering in general and in specific contexts. An interesting presentation provided an overview of threat modeling methods focusing on crowdsourcing the creation of Personae Non Gratae to model imaginary attackers.
RE for System of Systems
As a software ecosystem is, by nature, a system of systems, this topic also addresses software ecosystems. This year, REFSQ featured the first two papers addressing RE challenges in the context of software ecosystems. They addressed the modeling and analysis of openness trade-offs and how companies can define their contribution to open source software ecosystems. In addition, there was a paper on supporting runtime analysis of requirements violation in systems of systems. At RE2017, the focus was on identifying conflicting requirements in systems of systems and on defining business models in order to enable a system-of-systems approach to an underground mine.
To which of these oceans will you send your ships?