Data Spaces must preserve sovereignty and privacy while ensuring FAIR (Findable, Accessible, Interoperable and Reusable) principles. To do so, policy-based strategies have to be developed in order to describe the agreements reached in the Data Space. In this context, two open questions arise: how to define the right Data Space policies, as well as, how to enforce (and monitor) them. Despite the eforts towards defining and enforcing data access and usage policies, there is no solution to operationalize the enforcement of those considering data quality dimensions. However, data quality is becoming a hot topic due to the surge of federated learning and alternative analytical techniques, which require all providers to guarantee a data quality threshold in order to learn robust models. Currently, we have means to describe policies related to data quality rules (e.g., by combining standards such as ODRL and standard vocabularies) but we are missing means to elicit these policies from data providers and enforce them while preserving the data sovereignty. In this paper, we discuss the challenges and open questions that must be addressed in order to operationalize (and eventually, automate) data quality in Data Spaces, which span from requirements elicitation to data validation.
Data Spaces are federated ecosystems in which data
providers and consumers share data while preserving data
sovereignty and privacy. Currently, the Data Mesh
architecture [
Data Product concept, which provides a
productoriented view of the providers’ data assets. In short, the data
product is a node that encapsulates three structural
components required to function: code for enforcing policies (i.e.,
the Data Space agreements), data (and its metadata) and
infrastructure [
Behind the idea of Data Spaces is the objective of
extracting value from data sharing. This can be achieved in many
ways, but data analysis arises as prominent means to achieve
so, either by means of descriptive analysis (e.g.,
dashboarding and OLAP) or predictive analysis (e.g., learning models).
However, how to achieve data analysis in federated
environments is an open challenge, and federated learning [
One of the biggest open problems in Data Spaces not properly tackled is how the agreements reached (e.g., on guages and Analytical Processing of Big Data, co-located with EDBT/ICDT ∗Corresponding author.
LGOBE
https://futur.upc.edu/ClaudiaPatriciaAyalaMartinez (C. P. Ayala); https://futur.upc.edu/BesimBilalli (B. Bilalli); https://futur.upc.edu/CristinaGomezSeoane (C. Gómez); https://s.ua.es/_MuH (J. Mazón); https://futur.upc.edu/OscarRomeroMoral (O. Romero)
0000-0002-6262-3698 (C. P. Ayala); 0000-0002-0575-2389 (B. Bilalli); 0000-0002-3872-0439 (C. Gómez); 0000-0001-7924-0880 (J. Mazón); 0000-0001-6350-8328 (O. Romero) © 2025 Copyright for this paper by its authors. Use permitted under Creative Commons License
DQ) at the Data Space federated layer (i.e., at the federated
-unique- view of the data ecosystem) can be enforced at
the providers’ data assets regardless of their heterogeneity
and preserving data ownership and privacy. Note that this
problem has been easily tackled in centralized environments
by having a central authority extracting, transforming and
preparing data for analysis. However, this is not possible in
settings where data is not meant to be shared raw. For
example, the minimum number of instances and the variances
of key attributes might be set as DQ criteria for all data
providers and should be automatically and locally validated
by executing a software service (specific for the provider
infrastructure) provided by the Data Space services catalog.
The result of the service execution should be communicated
to the Data Space. To our knowledge, there is no
architecture, framework or solution tackling this problem, despite
the myriad of standards and definitions blooming around
the Data Space concept (e.g., [
We focus on how to validate DQ agreements in the Data Space and discuss the open challenges to make DQ happen in Data Spaces to enact trustworthy federated learning.
Data Spaces require a governance model for specifying DQ
agreements that stakeholders must adhere to in order to
participate. Importantly, this governance model must also
specify DQ needs agreed among data consumers and providers
when developing specific uses cases. Therefore, our view
is that the governance model for Data Spaces should
distinguish two levels: 1) a Data Space level for agreements
among stakeholders of the Data Space authority from data
regulations and strategic issues, and 2) a use case level for
agreements among data providers and consumers to build
specific Data Products. Based on this view and to facilitate
the discussion, we propose a visionary framework with a
process for the Data Space and use case levels (see Fig. 1).
Our framework follows the Open Data Product
specification [
ceur-ws.org ecutable part contains the machine-readable “as code” rules, provided as a service, to validate DQ dimensions. Next, we describe both processes and their main challenges.
DQ Requirements Engineering for Data Spaces.
Requirements engineering (RE) for complex systems in
open and dynamic environments that extend beyond a
single organization is widely recognized as a challenging
endeavor [
Extraction and Customization of DQ Rules. The complexity of DQ requirements and their textual or semistructured formalization make their direct operationalization challenging. With the aim of making DQ requirements executable in an operational environment, our visionary framework proposes to transform, in a semi-automated way and using specific catalogues for supporting this transformation, DQ requirements (at Data Space and use case levels) into formalized DQ rules that may be easily implemented. We propose to use a rule language with well-defined semantics (e.g., ODRL), to formalize DQ rules. Several challenges need to be tackled when performing this transformation: 1) the identification of relevant and suitable stakeholders with the specific knowledge for performing this activity in both levels; 2) the definition of specific catalogues with reusable transformation patterns for translating DQ requirements into rules, preserving their semantics; 3) the definition of the artifacts needed (e.g., specialized metamodels or new ODRL profiles), for automating the extraction and customization of DQ rules to the specific domain and level.
Implementation available as a Service of DQ Rules.
The inherent heterogeneity of providers in the context of Data Spaces renders the process of translating formal DQ rules into executable services a significant challenge. The main goal of this activity is to avoid building and maintaining custom solutions that are tightly coupled to specific execution environments or platforms. To address this, we propose an agnostic solution that leverages best practices from software engineering, such as containerized solutions, ensuring portability, scalability, and interoperability. However, the intrinsic characteristics of Data Spaces introduce several challenges that must be addressed: 1) dealing with heterogeneity at the infrastructure level by abstracting the diferences while ensuring consistent performance and security across environments; 2) allowing for dynamic and federated execution across multiple distributed nodes, ensuring real-time validation without requiring data centralization.
As conclusion, there is a need for further research to enact DQ in Data Spaces, a must for qualitative federated data analysis. In this sense, we have discussed a visionary framework, its main phases and challenges to be tackled.
This work has been partially supported by the EUHORIZON program under GA.101135513 (CYCLOPS) and by CIAICO/2022/019 project from Generalitat Valenciana.