Quality analysis

In line with the National Guidelines ("Linee guida nazionali per la valorizzazione del patrimonio informativo pubblico"), developed under the Data & Analytics Framework project by AgID and the Digital Transformation Team, we conducted a thorough quality assessment of our datasets to ensure their reliability and suitability for their intended purposes. Specifically, there are four main factors to look for when analysing data quality:

• Accuracy (syntactic and semantic): Verifying that the data and its attributes accurately reflect the real-world values or events they represent.
• Coherence: Ensuring that the data is consistent and free from contradictions when compared to other related datasets within the administrative context.
• Completeness: Assessing whether the datasets provide exhaustive values and fully account for all related entities (sources) that contribute to the defined procedures.
• Timeliness (or promptness of updating): Confirming that the data is up to date and corresponds to the relevant timeframes for the associated processes.

Legal analysis

The legal analysis of the source datasets is essential to ensure the long-term sustainability of the production process and of the publication of datasets and to guarantee a of both the data production process and the publication of datasets, while also guaranteeing a balanced service that aligns with public responsibilities and respects individual rights.

This analysis was carried out using a reference checklist consisting of a series of binary questions regarding the topics of: privacy issues, IPR policy, licenses, limitations on public access, economical conditions, and temporal aspects.

Publication license

A key aspect of the legal analysis is determining the appropriate publication license for the newly created mashup datasets. This decision must align with the licenses of the source datasets, which in our case, were all released under the CC BY 3.0.
To guide this process, we utilized the Licensing Assistant tool provided by the European Commission. After evaluating the options, we opted to publish all 10 mashup datasets under the CC BY 4.0 license.

The table below summarizes the original licenses of the source datasets and the final publication license applied to the mashup datasets:

Ethical analysis

For the ethical analysis of our project’s data, we applied the Data Ethics Principles and Guidelines and the Odi Project's detailed framework for assessing the ethical aspects of our data processing.

Both the source and mashup datasets for our project are exclusively derived from the Italian National Institute of Statistics (ISTAT). Therefore, we first focused on evaluating the fairness of ISTAT's data collection and management practices. Following that, we established clear ethical guidelines to ensure responsible handling of the datasets throughout our project's lifecycle.

Data Ethics Principles

• The "Open Voices" project aims to analyse the socio-cultural dynamics that influence violence against women in Italy, therefore the ethical approach in data management is fundamental to ensure that the rights of victims are respected, that no stereotypes or discrimination are fostered, and that the analysis contributes positively to raising awareness on such a sensitive issue.

  Human being at the center: ISTAT’s policy is deeply aligned with both ethical standards and legislative principles. The organization prioritizes the dissemination of statistical information to promote awareness of Italy’s social and economic conditions. It also strives to enhance public decision-making by providing clear, accessible statistical data. In addition, ISTAT conducts research to continually refine statistical methodologies and improve Italy's statistical literacy.
• Equality: In the context of the "Open Voices" project, the main objective is to analyze the socio-cultural dynamics that influence violence against women in Italy. Therefore, the collection of data concerns very sensitive issues such as gender violence and its perception within society. ISTAT data on key equality issues, such as gender discrimination and domestic violence.
• Transparency: ISTAT ensures transparency in data management by providing comprehensive documentation. This documentation covers the data collection methods, clarifies the use of specific terms and definitions, and outlines policies and licenses that safeguard against misinterpretation of the data.
• Accountability: ISTAT's quality assurance procedures align with European frameworks, specifically the European Statistics Code of Practice. This adherence strengthens both accountability and governance within the national statistical system and aligns it with European standards.
• Individual data protection: ISTAT ensures that its datasets are anonymized in compliance with legal requirements. The organization’s practices adhere to strict confidentiality standards, ensuring that the privacy of respondents is always respected. Their approach to handling sensitive data complies with European data protection laws (e.g., Regulation (EU) 2016/679 and Legislative Decree No. 322/1989).

Ethical concerns and their management

Although ISTAT adheres to ethical principles in data collection and management, the team has paid particular attention to the ethical management of source information, given the great sensitivity of the content dealt with in our project, which concerns gender-based violence. Data on sexual violence, the acceptability of partner violence and access to support services (such as 1522) are sensitive issues and ethical concerns in their treatment have been addressed in the following ways:

• Data Integrity and Privacy: To ensure data integrity and privacy, the values from the source datasets were aggregated and presented as percentages, avoiding any correlation with real individuals.
• Protection of Vulnerable Groups: Certain sensitive data were intentionally omitted to avoid the risk of discriminatory behavior. The project aimed to protect women experiencing violence, minimizing the risk of stigmatization.
• Avoiding Generalizations and Misinterpretations: The objective of the project was to identify potential patterns in the dynamics of gender-based violence, not to make inferences or generalizations. In our results and conclusions documentation, we emphasize that any observed patterns in the data should not be generalized due to inconsistencies in the data and the absence of other potentially relevant socio-economic factors.
• All relevant documentation regarding the data processing for the creation of mashup datasets and visualizations is provided in our GitHub repository.

Technical analysis

Source datasets

All source datasets have been assessed following the metadata model established by Agenzia per l'Italia Digitale (AGID), which categorizes metadata quality into four levels. This classification is based on two key factors: the strength of the data-metadata relationship and the level of detail provided.

1. Syntactic Quality: Ensures correct formatting and structural validity of metadata according to schemas and data models used.
2. Semantic Quality: Verifies the coherence and meaningfulness of metadata, ensuring alignment with controlled vocabularies or taxonomies.
3. Completeness: Assesses whether all mandatory and optional metadata elements are present to avoid missing critical information.
4. Consistency: Ensures logical consistency within and across datasets, verifying relationships and adherence to rules.

Note: Further details and reconstructed metadata for the source datasets are available in the metadata analysis table below.

RDF Metadata Assertion of the Datasets

All generated mashup datasets have been documented using metadata, adhering to the latest standard of DCAT Version 3 (August 2024). This choice reflects our commitment to leveraging the most up-to-date standards for metadata representation, ensuring flexibility and compatibility with modern technological frameworks.
While we referred to DCAT-AP_IT for alignment with Italian national guidelines, which are based on the earlier DCAT v1.0 (2016) and impose stricter constraints, our implementation adopted DCAT Version 3, as already stated. This approach was chosen to maintain interoperability within the national ecosystem while also benefiting from the advancements and enhanced features of the latest DCAT version of the standard. By doing so, our datasets align with the Italian Agency for Digitalization (AGID) directives, ensuring their integration with the broader public sector information (PSI) heritage, while also meeting the evolving needs of metadata management on an international scale.

Semantic Enrichment of Datasets

The metadata for the source datasets were primarily derived from the original data sources. When metadata was incomplete or unavailable, additional information was inferred and supplemented following the same principles applied to the mashup datasets. For instance, themes were assigned to source datasets based on recognized European Authority standards. To enhance the semantic description of datasets, we have utilized several key ontologies, including DCAT, DCTERMS, PROV, FOAF, ADMS, SKOS, CC, and DCAT-AP IT. These ontologies provide a structured framework for describing datasets and their metadata, ensuring interoperability and adherence to Linked Open Data (LOD) standards. This approach facilitates the discoverability, understandability, and reuse of datasets across diverse contexts.

The DCAT (Data Catalog Vocabulary) ontology plays a central role, serving to describe datasets and data catalogs and publish them on the Web. Key properties such as dcat3:dataset, associate datasets with a catalog, while dcat3:theme categorizes datasets by their topics, and dcat3:distribution specifies the available data formats. For example, datasets can be linked to themes like population statistics or societal issues using dcat3:theme. Learn more about DCAT Version 3 at W3C DCAT.

DCTERMS (Dublin Core Terms) is used to describe general metadata, including dcterms:title for dataset titles and descriptions, and dcterms:accessRights to indicate accessibility. This inclusion ensures the use of a widely recognized standard for metadata description. More information about DCTERMS can be found at Dublin Core Terms.

ADMS (Asset Description Metadata Schema) is included to complement DCAT by describing assets like datasets and services, particularly in government and public administration contexts. ADMS plays an important role in the asset management landscape. For further details, consult the ADMS Specification.

CC (Creative Commons) provides a vocabulary for licensing datasets, such as cc:license, ensuring clear and standardized license attribution. Visit Creative Commons specifications for further insights.

Finally, DCAT-AP IT (DCAT Application Profile for Italy) is referenced through dcatapit: to align with Italian government open data standards. It extends DCAT to meet specific national requirements. For more information, check out DCAT-AP IT.

FAIR principles

During the project development, we aimed to align our efforts with the FAIR principles established by the GO FAIR Initiative.
These principles, developed by a consortium of scientists and organizations, provide guidelines to ensure digital assets are Findability, Accessibility, Interoperability, Reusability, with a strong emphasis on machine-actionability.

The table below utilizes the FAIR Principles overview from GO FAIR as a checklist to systematically evaluate our project’s adherence to these guidelines.