This year we published a paper in the industrial track of ER 2011 [1], the annual conference on conceptual modeling research. In this context I spent a couple of days in Brussels at the conference. This gives me the opportunity to share with you an (high level) overview of the trends in this area. Please, notice that this post is only a summary of the discussions I saw and talks I had the chance to see.
Ontology and design
There was a significant number of papers presenting the usage of ontologies for having a more meaningful and accurate design. There was applications for better desing validation, for better understanding of the design by all stakeholders including customers, and even for a better and more accurate generation from design to operational model.
Concerning the ontologies, there was an interesting debate during the conference initiated by Antoni Olivé about the need and the feasibility to build an unified universal ontology. This interesting question directly led to a set of philosophical and interesting questions such as how can we define the limit of a concept especially from different cultures. The example that came back several time was about the concept of a person, how to limit it? How can we agree on an unified definition of a person. As you can guess the debate remains open.
Variability
This year, there was a special Variability track. For those of you for who the variability looks a mysterious term, here a very short definition. The variability is a new discipline coming from the Software product line [2], an new approach for designing flexible software and supporting mass customizations. There is mainly 2 parts: the (1) domain platform which is generic and provides variability points and (2) application domain that extend those variability point for implementing variations.
At ER we had the chance to see one of the main architect of the SPL standard (ISO 2650 -> 2656) initiative presenting the content and the state of the process. Several talks showed the importance of the variability for better design and described different use cases and usages.
Close to research and Close to business
We can definitely say that the organizers of the conference focused on bridging the business (the most important consumers of modeling in BI and DW) and the researches. We can cite among others the KPI management in BI, DW design for business managers or even security aspects.
Visual Design
The visual design in conceptual modeling has always been an important topic as the visual representation translates the underlying model. This year several talks showed the important of this discipline especially for the impact on the modeling, complexity management as the models becomes richer and more complicated and finally expressiveness.
Something missing?
I have to say that I have been disappointed to see very few things about big data. The only mention was made by Sudha Ram, from the University of Arizona, in a panel discussions about new directions. She explained that if we have a huge number of data, a first and mandatory step for data integration is to link them by a semantic layer. In discussion with Prof. Ram, we arrived at the conclusion that adding a semantic layer will make us end up with a huge graph. Therefore the current modeling approach for BI are de-facto not adapted. An interesting research area would be the re-think of the BI design approach such as the *-schema or the snowflake schema, but suited for graphs or even to distributed data and to be able to generate an efficient operational model to deal with those particularities.
References
[1] 30th International Conference on Conceptual Modeling (ER 2011), Brussels, Belgium , http://er2011.ulb.ac.be/
[2] K. Pohl and al., Software Product Line Engineering: Foundations, Principles and Techniques. Springer, 2005