10. – 15. April 2016, Dagstuhl Perspectives Workshop 16151
Foundations of Data Management
Marcelo Arenas (Pontificia Universidad Catolica de Chile, CL)
Richard Hull (IBM TJ Watson Research Center – Yorktown Heights, US)
Wim Martens (Universität Bayreuth, DE)
Tova Milo (Tel Aviv University, IL)
Thomas Schwentick (TU Dortmund, DE)
1 / 2 >
Auskunft zu diesem Dagstuhl Perspectives Workshop erteilt
The focus of Foundations of Data Management (traditionally termed Database Theory) is to provide the many facets of data management with solid and robust mathematical foundations. The field has a long and successful history and has already grown far beyond its traditional scope since the advent of the Web.
The recent push towards Big Data, including structured, unstructured and multi-media data, is transforming and expanding the field at an unusually rapid pace. However, for understanding numerous aspects of Big Data, a robust research exploration into the principled foundations is still lacking. This transformation will call upon the Database Theory community to substantially expand its body of tools, techniques, and focal questions and to much more fully embrace several other disciplines, most notably statistics and probability theory, natural language processing, data analytics, emerging hardware and software supports for computation, and data privacy and security.
Big Data is not the only force that is driving expansion and transformation for the Foundations of Data Management. With the increasing digitization of diverse industries, including "smarter cities", education, healthcare, agriculture and others, many diverse kinds of data usage at large scales are becoming crucial. The push towards data-centric business processes, which are especially important for knowledge-worker driven processes, raise fundamentally new questions at the intersection of data and process. And increasing adoption of semantic web and other ontology-based approaches for managing and using meta-data push the boundaries of traditional Knowledge Representation.
The purpose of this Dagstuhl Perspectives Workshop was to explore the degree to which principled foundations are crucial to the long-term success and effectiveness of the new generation of data management paradigms and applications, and to understand what forms of research need to be pursued to develop and advance these foundations.
For this workshop we brought together specialists from the existing database theory community, and from adjoining areas, such as Machine Learning, Database Systems, Knowledge Representation, and Business Process Management, to understand the challenge areas that might be resolved through principled foundations and mathematical theory.
More specifically, during this workshop we worked on:
- Identifying areas, topics and research challenges for Foundations of Data Management in the forthcoming years, in particular, areas that have not been considered as Database Theory before but will be relevant in the future and of which we expect to have papers at PODS and ICDT, the main conferences in the field.
- Outlining the techniques that will be most fruitful as starting points for addressing the new foundational challenges in Data Management.
- Characterising the major challenge areas in Big Data where a principled, mathematically-based approach can provide important contributions.
- Finding research goals in neighbouring areas that may generate synergies with our own.
The workshop consisted of eight invited tutorials on selected topics:
- Managing Data at Scale,
- Uncertainty and Statistics in Foundations of Data Management,
- Human in the Loop in Data Management,
- Machine Learning and Data Management,
- Data-Centric Business Processes and Workflows,
- Ethical Issues in Data Management,
- Knowledge Representation, Ontologies, and Semantic Web, and
- Classical DB Questions on New Kind of Data. The abstracts of these talks can be found below in the document.
There were also seven working groups on theory-related topics, which identified the most relevant research challenges for Foundations of Data Management in the forthcoming years, outlined the mathematical techniques required to tackle such problems, and singled out specific topics for insertion in a curriculum for the area. The topics of these working groups were:
- Imprecise Data,
- Unstructured and Semi-structured Data,
- Process and Data,
- Data Management at Scale,
- Data Management and Machine Learning,
- Knowledge-Enriched Data Management, and
- Theory and Society.
There was also a working group on curriculum related issues, that collected and enriched the information provided by the working groups about the design of a curriculum on Foundations of Data Management. Each one of these groups worked for two consecutive hours in different days. Workshop participants had to participate in at least two working groups, although most of the people participated in four of them. Summaries of the discussions held in each one of these working groups can be found below in the document.
During the first day of the workshop, there were also five working groups that analysed several community-related aspects. In particular:
- Attraction of women and young members,
- cross-fertilization with neighbouring areas,
- relationship to industry,
- impact of our research, and
- the publishing process.
The discussion within some of these working groups gave rise to the creation of specific tasks to be accomplished by the community in the following years. These tasks will be coordinated by the councils of PODS and ICDT, the two main conferences in the field.
This Dagstuhl Report will be accompanied by a Dagstuhl Manifesto, in which the outcome of the different working groups will be explained in more detail and several strategies for the development of our field will be proposed.
Creative Commons BY 3.0 Unported license
Marcelo Arenas and Richard Hull and Wim Martens and Tova Milo and Thomas Schwentick
- Data Bases / Information Retrieval
- World Wide Web / Internet
- Data Management
- Database Theory
- Big Data
- Machine Learning
- Data Mining
- Information Extraction
- Graph Algorithms
- Knowledge Representation
- Semantic Web
- Scientific Workflow
- Data-centric Workflow
- Probabalistic Data