Achieving Interoperability in Dataspaces

Breaking Down Data Silos

Data is a highly valuable asset in the digital economy, but its potential value can only be realized if it can move and interact with other data to produce insights that create value. For this it must be possible for data to be shared and reused in a trusted way. Dataspaces are a new architecture paradigm for data ecosystems that promises to resolve the trust issue between data sharing parties.

For data to flow we need Interoperability, which is the ability of different systems and organizations to exchange, understand and use data. It is essential to enable trusted data sharing and creating value in data ecosystems.

Dataspaces help to establish a common understanding of trust, provide a mechanism to establish sharing contracts, which include access and usage policies that ensure the protection and accountability of data providers and data consumers.

But what is a dataspace exactly?

It is a logical collection of participants (organizations, or legal persons) that share a common context or domain and offer data sharing contracts and capabilities to access their data assets in a trusted way to other participants. Different dataspaces may have different goals, architectures, business models, and governance structures, depending on the authority or community that drives them. To avoid fragmentation and duplication of efforts, participants in these dataspaces need to communicate in an interoperable way with each other and within multiple dataspaces, following common standards and principles.

The guiding principles of a dataspace are:

1. Self-determined control of data use (Data Sovereignty) is of utmost importance and should be the ideal vision that each dataspace thrives to enable
2. Participants have autonomy and are able to act with choice
3. Participants have agency over their data assets
4. A dataspace creates a context of trust
5. The Dataspace Governance Authority (DGA) is a governance body for a dataspace

The DGA is a logical function in the dataspace and while it will be quite common to combine the DGA with the legal organization of the dataspace, it is also possible that a DGA exists without any legal organization operating it. (A DGA could be just a set of policies passed around between dataspace participants without any single owner, just being agreed on by a consensus algorithm between participants.)

The DGA is also responsible for the semantic models of the dataspace and thus has a huge influence on the interoperability at that layer.

Taking the guiding principles above into account leads us to the conclusion that interoperability is a shared responsibility between the participants and the DGA.

The more autonomy and agency a participant can act with, the more responsibility for ensuring interoperability layers with the participant. With less autonomy and agency, more interoperability responsibility moves to the DGA and legal organization layers thus lessening the burden of interoperability on the participant. 

Autonomy and agency of participants also comes with increased responsibilities for the participants.

Facets of interoperability

Interoperability can be achieved at different levels, depending on the degree of integration and alignment of the data and systems involved. Two well-known frameworks that define interoperability levels are the ISO/IEC 19941 standard for cloud computing interoperability and portability, and the European Interoperability Framework for public services. Both frameworks identify four main levels of interoperability: technical (transport & syntactic), semantic, organizational, and legal:

• Technical interoperability refers to the physical and logical connections between systems and data sources, such as protocols, interfaces, and formats. This includes syntactic interoperability which refers to the structure and syntax of the data exchanged, such as schemas, models, and vocabularies.
• Semantic interoperability refers to the meaning and interpretation of the data, such as concepts, relationships, and ontologies.
• Organizational interoperability refers to the processes, policies, and governance of data sharing, such as roles, responsibilities, and agreements.
• Legal interoperability refers to the acceptance of legal equivalence of contracts and contractual clauses between different participants and also between different data ecosystems. These ecosystems can have differences on multiple dimensions,  based for example on industry regulations, or national laws but also contractual statements with identical wordings might have diverging interpretations in different data ecosystems.

Interoperability Models

When talking about interoperability in dataspaces we need to separate the discussion between two main interoperability models:

1. Intra-Dataspace: interoperability within a dataspace between individual participants (and also between the participants and the DGA of that dataspace)
2. Cross-Dataspace: interoperability where a participant wants to access data which is being shared through multiple dataspaces

Intra-Dataspace interoperability is about the interoperability within a dataspace. This focuses on how participants interact with each other, and as well as with the DGA. The DGA defines what rules govern the dataspace. This includes also which version of the Dataspace Protocol needs to be used, what identity protocols and standard to use, which Trust Frameworks are accepted, what semantic models need to be understood and so on. Participants have the responsibility to at least support and understand the protocols and models that the DGA mandates but can also support additional versions and semantic models.

Cross-Dataspace Interoperability refers to the interoperability required for one entity to participate in multiple dataspaces. As it is the participant that wants to access data from two different dataspaces most of the responsibility for interoperability falls on the participant.

A participant needs to become a member of both dataspaces, thus fulfilling the membership rules to be able to join both dataspaces. This implies that the participant is able to support all the protocols and semantic models that both dataspaces require. Should those not be identical it is up to the participant to be able to support the right protocols and their version in each dataspace and potentially do any necessary mappings. Another option is when the DGA, as well as the legal entity operating the dataspace (if such exists) can support participants by agreeing with other DGAs and legal entities from other dataspaces on supported protocols and semantic models. This can greatly reduce the burden on the participants in sharing data to and using data from multiple dataspaces.

Trust Frameworks and Trust Anchors

As the DGA also defines which Trust Frameworks and which Trust Anchors can be used by participants within a dataspace all the before mentioned interoperability facets also apply to Trust Frameworks and Trust Anchors. As it is very likely that Trust Frameworks and Trust Anchors will support multiple dataspaces it is especially important that the applicable protocol versions and semantic models are clearly defined.

Improving Interoperability in dataspaces

As already established above, the main responsibility for interoperability in dataspaces is with the participant, however, everyone involved in a dataspace can support interoperability by aligning with other parties:

Dataspace to dataspace: negotiate legal equivalency of processes and rules between the two dataspaces. This is many times driven by not-for-profit or industry consortia.

Dataspace to Trust Frameworks/Trust Anchors: Align on mapping between policies and legal provisions and processes. If multiple regulatory and legal areas are covered, some help by regulators and law makers might be needed to establish the right mapping and finding solutions within the boundaries of national legal systems and regulatory processes.

Dataspace to DGAs: Align on governance models and organizational processes.

Trust Framework to other Trust Frameworks: Share semantic models for policies and align on claims exchange protocols and profiles is required.

Trust Framework to DGAs: Agree on standardized claims exchange protocols and a common set of semantic models.

DGA to other DGAs: Share semantic models for policies and agree on functional processes.

If you want to learn more about work that is being done to enable the interoperability of dataspaces, please, visit the Eclipse Dataspace Working Group and join the mailing list!

Disclaimer: This article is based on the text contributed under CC-BY to the IDSA Rulebook where it has been refined by the dataspace community. Please, check out the chapter on Interoperability there for even more great insights and details!