Inter-Organizational Data Sharing: From Goals to Policies to Code
By Matthias Jarke
iStar - MREBA Joint Keynote @ER2019
He presented a very interesting table showing the business ecosystems issues that the IDS initiative aims at addressing:
----
He presented some requirements w.r.t data sharing in business ecosystems (Otto/Jarke 2019).
----
Dependency Structure in Collaborative Data Sharing generate some obstacles:
- hierarchical or competitive behavior drive people to deny sharing data.
- sometimes, data may be only used for some processes (e.g. for management, but not for purchasing and sales)
- sometimes, data must be deleted within a certain period of time.
----
From Access Control to Data Usage Control - The data usage goals are not achievable by Access Control alone.
These goals are:
Secrecy
Integrity
Time to live
Anonymize (aggregation)
Anonymize (replacement)
Separation of duties
Usage scope (application)
He showed a model called XACML - Access Control Data Flow Model
*it seems very interesting
----
There must be a gradual automation and regulation and contracts for Data Sharing. This is the only way to guarantee that it is going to work effectively.
----
Different kinds of technologies may be used to alleviate the delays of the different policies (e.g. secrecy, integrity etc.). These may really slow down the database system. Thus, technologies like annotation-based systems that will provide quick response instead of delays may be used, so as to optimize the process of data access.
----
Elaborated Role-Dependency Structure in the International Data Spaces (IDS) Reference Architecture Model (Otto 2019)
This work is based on a more elaborate kind of i* model.
----
Semantic Interoperability
Mapping Specification
- Metamodeling approach - deductive database formalisms enhanced by (second-order) TGDS (CLIO), or Telos multi-level language enhanced by SO-TGDS and factored by roles (GeRoMe)
The first metamodeling techniques were not so sophisticated as the modern ones. You must be careful not to get to undecidable scenarios, as the model gets more complex.
- Ontology-based mapping - DL based approaches (Lenzerini et al.) or RDF-based probabilistic knowledge graph schemes.
Mapping Creation and Analysis
- Mapping Discovery (see talk by Rihan Hai at ER 2019)
- Analysis (consistency, completeness)
- Model-managed
---
Semantic Perspective on Data Sharing (Lenzerini 2019): Mapping Between (Pre-defined or Mined) Schemes.
*This is a must-read reference
---
Outline of work in this topic
Requirements
Recently a number of alliance driven platforms developed in contrast to typical keystone driven platform.
Companies feel they must collaborate in these ecosystems and they develop joint platforms that put them in the center of the ecosystem to which they belong.
The IDS initiative may also contribute to a digital infrastructure by addressing its data layer.
Virtual Hyperscaler for Obtaining Performance in Alliance-driven Data Ecosystems.
----
Q&A
Eric: in which areas can iStar contribute more?
Answer: This area includes an agent dependency point of view, which may be an opportunity for the use of i*; Frank Pillar, partner in the project, is interested in conceptual modeling to solve some of the IDS problems.
Giancarlo: are there connections from IDS with Fair Data project, since these projects share some commonalities and goals?
Answer: yes. He sees that IDS is focused on general data, but some of the data is research data and thus, he has a couple of PhD students working on Fair Data related stuff.
Wolfgang: How can users feel better about sending their data, and avoid the feeling of loosing control over the data?
Answer: The idea is to distribute only part of the data (e.g. anonymizing it). Other requirements, such as separation of duties may also contribute to this; avoiding hierarchical structures
By Matthias Jarke
iStar - MREBA Joint Keynote @ER2019
He presented a very interesting table showing the business ecosystems issues that the IDS initiative aims at addressing:
----
He presented some requirements w.r.t data sharing in business ecosystems (Otto/Jarke 2019).
----
Dependency Structure in Collaborative Data Sharing generate some obstacles:
- hierarchical or competitive behavior drive people to deny sharing data.
- sometimes, data may be only used for some processes (e.g. for management, but not for purchasing and sales)
- sometimes, data must be deleted within a certain period of time.
----
From Access Control to Data Usage Control - The data usage goals are not achievable by Access Control alone.
These goals are:
Secrecy
Integrity
Time to live
Anonymize (aggregation)
Anonymize (replacement)
Separation of duties
Usage scope (application)
He showed a model called XACML - Access Control Data Flow Model
*it seems very interesting
----
There must be a gradual automation and regulation and contracts for Data Sharing. This is the only way to guarantee that it is going to work effectively.
----
Different kinds of technologies may be used to alleviate the delays of the different policies (e.g. secrecy, integrity etc.). These may really slow down the database system. Thus, technologies like annotation-based systems that will provide quick response instead of delays may be used, so as to optimize the process of data access.
----
Elaborated Role-Dependency Structure in the International Data Spaces (IDS) Reference Architecture Model (Otto 2019)
This work is based on a more elaborate kind of i* model.
----
Semantic Interoperability
Mapping Specification
- Metamodeling approach - deductive database formalisms enhanced by (second-order) TGDS (CLIO), or Telos multi-level language enhanced by SO-TGDS and factored by roles (GeRoMe)
The first metamodeling techniques were not so sophisticated as the modern ones. You must be careful not to get to undecidable scenarios, as the model gets more complex.
- Ontology-based mapping - DL based approaches (Lenzerini et al.) or RDF-based probabilistic knowledge graph schemes.
Mapping Creation and Analysis
- Mapping Discovery (see talk by Rihan Hai at ER 2019)
- Analysis (consistency, completeness)
- Model-managed
---
Semantic Perspective on Data Sharing (Lenzerini 2019): Mapping Between (Pre-defined or Mined) Schemes.
*This is a must-read reference
---
Outline of work in this topic
Requirements
- Trust among members
- Data sovereignty - who owns the data?
- Shared governance
- Interoperability
- Compliance with antitrust legislation
- Data economics - money sharing while data is shared.
Recently a number of alliance driven platforms developed in contrast to typical keystone driven platform.
Companies feel they must collaborate in these ecosystems and they develop joint platforms that put them in the center of the ecosystem to which they belong.
The IDS initiative may also contribute to a digital infrastructure by addressing its data layer.
Virtual Hyperscaler for Obtaining Performance in Alliance-driven Data Ecosystems.
----
Q&A
Eric: in which areas can iStar contribute more?
Answer: This area includes an agent dependency point of view, which may be an opportunity for the use of i*; Frank Pillar, partner in the project, is interested in conceptual modeling to solve some of the IDS problems.
Giancarlo: are there connections from IDS with Fair Data project, since these projects share some commonalities and goals?
Answer: yes. He sees that IDS is focused on general data, but some of the data is research data and thus, he has a couple of PhD students working on Fair Data related stuff.
Wolfgang: How can users feel better about sending their data, and avoid the feeling of loosing control over the data?
Answer: The idea is to distribute only part of the data (e.g. anonymizing it). Other requirements, such as separation of duties may also contribute to this; avoiding hierarchical structures
