RDA 11th Plenary - Programme

Learn about professional data/metadata management at large scale.

The IEEE Big Data Governance and Metadata Management group (BDGMM) will run a workshop and hackathon along with the Research Data Alliance (RDA) Plenary 11. Governance and metadata management poses unique challenges with regard to the Big Data paradigm shift. It is critical to develop interoperable data infrastructure for Big Data Governance and Metadata Management that is scalable and can enable the FAIR (Findability, Accessibility, Interoperability, and Reusability) principles between datasets from various domains without worrying about data source and structure. The goal of this multidisciplinary workshop is to gather both researchers and practitioners to discuss methodological, technical and standard aspects for Big Data management. Papers describing original research on both theoretical and practical aspects of metadata for Big Data management are solicited. CLARIN offers its huge metadata space (800.000 records harvested from providers around the world). For more information see https://bigdatawg.nist.gov/bdgmm2018.html.

Organisers/Presenters: Wo Chang (NIST, Chairman of the IEEE BDGMM group), Priyaa Thavasimani (Newcastle University, member of the IEEE BDGMM group), Dieter van Uytvanck (Technical Director CLARIN infrastructure)

Learn about Machine Learning methodology and typical industry use cases! Ever increasing data amounts and their heterogeneity with respect to types require new methods to detect hidden patterns in aggregated data sets. Machine Learning is one of these new promising methods having the attractive characteristic of requiring only a limited amount of concrete specification, relying on the data to drive and infer the models. Leading experts will give an introduction to the basic methods and refer to software libraries that can be used. Contributions about concrete industry use cases. The session is meant to give a view about how these methods can be used and what the challenges are when using them.

• Introduction:
  • Mikio Braun (Zalando): Introduction to ML and Advanced Methods
• Presentation of Industry Use Cases:
  • Wojciech Samek (Fraunhofer HHI): Opening the Black Box of Deep Learning
  • Hassan Chafi (Oracle): Graph Powered ML and Analytics: An Overview and Use Cases.
  • Andreas Klug (ITyX AG): How AI will change the CX paradigm (Customer Experience)
  • Aylin Koca (UnternehmerTUM): Starting Up

Learn about the challenges and solutions to work with real-time data.

When industry talks about IoT-data they often mean real-time data provided for example by sensors, industry controls or cars and real-time access to these data. Many research data are also generated by machines but most of them are usually not accessible in real-time. On the other hand, very few research and scientific institutions are prepared to provide such data including real-time access. In a conversation between a research data provider, a company representative and the audience, we want to discuss ideas for working with IoT (real-time) data. What are the obstacles and where are common interests? How can a mutually beneficial cooperation be achieved? Does IoT imply new challenges for research data and their provisioning?

Again we received a number of quite interesting sketches of new business cases where data is being collected and aggregated to give services that have a commercial relevance. While "localexpert24" is clearly directed to "mittelstands" economy to increase efficiency and reduce errors in daily work, the other two are inspired by research work, but indicate substantial economic potential. It is just per accident that two speakers have the same name, but they come from completely different backgrounds and locations. For more information see http://ideenwettbewerb.mpcdf.mpg.de

Meet the Experts on Major Data Challenges

In this interactive session important questions for potential entrepreneurs will be addressed. Well-known key persons will be available to give their views on the emerging data economy, give advice how to initiate a start-up and how to guide an SME to successful data businesses, and answer questions. Various issues of broader interest such as rights on data, liability conditions, usage licenses, and cross-national aspects, but also stability and availability of services, roles in the data landscape, etc. will be addressed.

Presenters:

• tbc (Bitkom): Bitkom views and experiences
• Wolfgang Dorst (former Bitkom): Questions about chances and risks of the data economy
• Georg Wittenburg (Inspirient): Challenges and Opportunities for Startups
• Nicolas Zimmer (Technologie Stiftung Berlin): How to define useful business cases
• Mark Hahnel (Figshare): Global Research Data - Building infrastructure to defeat policy clash
• Jian Zhou (Allied for Startups): Growing Data Economy in China
• Ulrich Schwardtmann (GWDG): Generic Data Management Services in Science and Business

Learn about professional data/metadata management at large scale.

The IEEE Big Data Governance and Metadata Management group (BDGMM) will run a workshop and hackathon along with the Research Data Alliance (RDA) Plenary 11. Governance and metadata management poses unique challenges with regard to the Big Data paradigm shift. It is critical to develop interoperable data infrastructure for Big Data Governance and Metadata Management that is scalable and can enable the FAIR (Findability, Accessibility, Interoperability, and Reusability) principles between datasets from various domains without worrying about data source and structure. The goal of this multidisciplinary workshop is to gather both researchers and practitioners to discuss methodological, technical and standard aspects for Big Data management. Papers describing original research on both theoretical and practical aspects of metadata for Big Data management are solicited. CLARIN offers its huge metadata space (800.000 records harvested from providers around the world).

Organisers/Presenters: Wo Chang (NIST, Chairman of the IEEE BDGMM group), Priyaa Thavasimani (Newcastle University, member of the IEEE BDGMM group), Dieter van Uytvanck (Technical Director CLARIN infrastructure)

Experts tend to agree that the culture with respect to data trading/sharing needs to change towards a domain where data is seen as any other good. This will require a clarification of the conditions (re-usage, liability, etc.) to improve data sharing. We need to understand that there needs to be a separation between data creators, brokers and users. Currently too many data creators keep their data closed and thus inhibit new services and innovation. Unclear rights situation may stimulate such a mentality. A number of questions should be discussed by this panel: How do we need to shape ownership, copyright and licenses to make the data economy working? Which type of data should be available to foster broad innovation? How can we leverage the unique strengths of the European market to build a globally competitive data economy?

Along with the clarification of sharing/trading conditions we can assume that the data culture industry will change stepwise and foster the emergence of the global exchange of data and the European Digital Single Market. The European continent is following a dual evolution: on the one hand supporting the creation of an integrated European Digital Single Market, and on the other hand simultaneously substantially strengthening the privacy with the adoption of the European General Data Protection Regulation. This apparent paradox is just one example highlighting the problems we are confronted with to come to a flourishing data economy. Regulations with respect to data sharing in different countries or regions are rather diverging and it is not obvious whether this will hamper data sharing/trading or not. Establishing a global data market seems to be complex requiring at the end brokers who can deal with all the different regulations and give advice. The session will explore the consequences of the above mentioned paradox by highlighting emerging trends and approaches to conciliate these distinct but complementary sets of requirements and will put this into a global perspective. Which approach should be taken by companies to participate in the data market given the uncertainties? Is a global data market an illusion?

As long as projects or initiatives work in their own "silo" domain where strict guidelines can be enforced, data management and processing may not be optimally solved, but clear architectural choices guarantee interoperability to a certain extent. However, when data from different silos are brought together we are confronted with severe interoperability challenges and quality differences. Therefore, in industry experts are now starting to talk about a transition from the traditional approach of creating structured and unified "data warehouse" to the much more open approach of "data lakes" to meet the challenges of the emerging data domain.

The trends are clear: For discovering the richness of data and create new business cases it will increasingly be important to combine data from different sources. We can observe a few strategies to achieve both operational efficiency as well as a leading, defensible market position: 1) Some companies feel so strong and important in market sectors that they hope to be able to define "platforms" which others need to adapt to. How to design such platforms to be able to influence market sectors is not yet clear. 2) Some companies are relying on specifying generic reference architectures based on a holistic analysis of the grand problem and then follow a top down design process by stepwise concretisation down to components which are then implemented and tested. 3) Some are still waiting on making decisions by observing the high dynamics in the data domain where bottom-up mechanisms such as driven by RDA and other initiatives are playing the major driving force. A number of questions should be taken up by this panel: Which approaches should we follow to come to durable data infrastructures and save investments? Which kind of interactions do we need to make progress in increasing interoperability? Will enforced platforms be sustainable or will open best practices be the way to go? How can we learn from the many experiments and how much time do we have?

In these discussions about the most promising approaches to come to durable components of future data infrastructures we do not start from scratch. Much experience has been gained within the last decade mainly with a number of components and technologies. As example, we can refer to the wide range of semantic technologies which have been developed that yet still have a limited relevance for the daily data practices. Also we can refer to architectures around the concept of digital objects which was at the base of cloud system designs, but also resulted in clear ideas about persistently and uniquely defining identities and building a globally available resolution system. A number of questions should be taken up by this panel: Which components/technologies are unmissable and have a potential for broad uptake? What is missing for some of these components/technologies hampering their uptake in industry practices? What needs to be done to make them being accepted and thus ensure investments?