ER 2020

Conceptual Modeling in the Face of Complex Social Realities

Eric Yu, University of Toronto, Canada

It has long been recognized that the effectiveness and viability of information technology systems are not determined by technology alone. The first part of this talk will retrace the development of the i* framework – an attempt to use Conceptual Modeling to bridge a social conception of the world and the design of information systems. i* adopts a rather simplistic view of the social world. It consists of actors who depend on each other to achieve what they want. Actors have freedom to choose among alternative means for achieving their goals, but their choices are constrained by mutual dependency relationships among actors. As technology options present themselves, actors seek to reconfigure relationships in ways that would advance their strategic interests.

By applying i* modeling, one can explore the potential impacts of various system designs on stakeholders, in search of design options that would better meet the desires and aspirations of all concerned. i*-inspired social modeling has seen applications in business process analysis, requirements engineering, software development methodologies, information security and privacy, and other areas. Many researchers have incorporated social modeling into their modeling methods and techniques, and have enhanced and extended the original proposal in different directions. A version of i* is part of an international standard.

In the second part of this talk, I will suggest some open research challenges for Conceptual Modeling. Today, interactions between technology and social realities have become ever more intricate and intense. Data-driven applications are complementing and sometimes displacing human knowledge work. Software systems are sensing and manipulating human emotions – on social media platforms, in gamified applications, and in affective computing. Personal and public health, environmental responses, and citizen participation are all mediated through data and information systems. As systems become embroiled in these complex social and technological realities, the simplified ontology of i* may seem woefully inadequate.

Conceptual Modeling to date has served as the core foundation for building data-based applications. Can Conceptual Modeling also provide the foundation for conceptualizing and designing socially complex applications?

Biography:

Eric Yu is Professor at the Faculty of Information at the University of Toronto. His research interests include conceptual modeling, software requirements engineering, information systems engineering, knowledge management, and enterprise modeling. In his PhD work, he developed the i* framework for social modeling. The work has inspired hundreds of research papers, dozens of PhD theses, and many software tools. A version of i* is part of an international standard. He is co-editor of the MIT Press book series on Information Systems, and is on the editorial boards of the Requirements Engineering journal, IET Software, and the Journal on Data Semantics. He was Program Co- Chair for ER 2008 and 2014, and for CAiSE 2020. He was recipient of the 2019 Peter P. Chen Award.

Tool awareness in software development: A reflection on the relevance of Conceptual Modeling and Machine Learning

Christiane Floyd, TU Wien

The discipline of Software Engineering, founded in 1968 as a result of the so-called Software Crisis, can be characterized as history of hypes revolving around successive or overlapping methods and approaches, each promising to reduce or even eliminate persistent problems in software development.

Meanwhile, software engineers have learned to assess and appreciate these approaches as tools – relevant for a limited range of applications, coming with their own agenda, and creating new troubles while removing existing ones. Tool awareness (a term coined by Peter Naur in the 1980s) designates the ability to understand how tools work and can be used effectively, to choose appropriate tools, tailor them to the tasks at hand and combine them with other tools, in order to manage the overall process effectively. From this perspective, both conceptual modeling and machine learning are to be regarded as tools in software development.

Although conceptual modeling is a well-established approach which in certain areas of software development can be regarded as a standard, some of its claims, e.g. automatic code generation, are controversial. It remains a challenge to delineate the class of systems and the parts of software development for which conceptual modeling is a key approach, how it relates to other approaches (such as architecture design) and how models can remain relevant for software systems over time, given the inevitable changes in systems.

By contrast, machine learning and search-based techniques are relatively new approaches with little experience available. They are mostly advocated as aid to reduce cost and speed up certain parts of the process, however their place in the overall process is not yet well understood. For which types of systems are these approaches relevant? How do they relate to basic concerns in Software Engineering, such as: Communication with users? And within the development team? Understanding code and being able to correct it, in case of errors? To what extent can software features, acquired by machine learning be explained and made transparent? How can the need to optimize certain features by applying ML techniques be combined with other quality considerations, such as the ability to make required changes?

The dialogue between the software engineering community and the proponents of new approaches should not be based on grand claims but on transparent criteria and on experimental application. Rather than promoting rivalries between competing schools of thought, the center of discussion should be the practice of software development, which is becoming one of the key occupations for social development world-wide.

Biography:

Christiane Floyd is professor emerita for software engineering and honorary professor at the Technical University of Vienna. As head of the software engineering group at the Technical University of Berlin (1978 – 1991) and the University of Hamburg (1991- 2008), she was the principal author of STEPS, a participatory and eolutionary approach to software development, and explored epistemological foundations for cooperative knowledge projects. Since 2006 she is committed to promoting the use of information and communication technologies for development in Ethiopia.

Modelling the Future Climate – Concepts and Challenges

Thomas Ludwig, German Climate Computing Center (DKRZ), Germany

Climate modelling and projecting future climate scenarios is a complex task that requires computers of highest performance. The models are represented by computer programs that often comprise several hundreds of thousands of lines. Their execution produces huge volumes of data that need to be analyzed and visualized. The existing ecosystem of computational climate modelling is now confronted with new challenges: machine learning incorporates new algorithmic methods, the end of Moore´s Law limits the performance of future supercomputers and the ever rising power consumption of the systems sets new limits to knowledge gaining. The presentation will give an overview over the development of computational climate modelling and how the discipline will tackle upcoming challenges.

Biography:

Prof. Dr. Thomas Ludwig studied informatics at Universität Erlangen-Nürnberg, Germany. He received his PhD and also habilitation degree from Technische Universität München. From 2001 to 2009 he had a professor position at the Universität Heidelberg. Since 2009 he is professor for scientific computing at the Universität Hamburg. He is also the director of the German Climate Computing Center (DKRZ). His research focus is on high performance computing and in particular on storage system.