Provenance-rich modelling improves auditability
Representing contributions, evidence, procedures, and decisions with explicit provenance lets participation flows be inspected and reused more reliably than platform-specific or document-only records.
PhD Thesis Project · Informatics Engineering
Linked Open Democracy
The Digital Contract for Auditable Civic Participation in Europe.
Author: Paulo Jorge Pereira Martins
Supervisor: Prof. Hugo Gonçalo Oliveira
University of Coimbra · Department of Informatics Engineering · CISUC
Digital Democracy
Knowledge Graphs
Semantic Web
NLP
Provenance
Trustworthy AI
The project
Trustworthy participation, kept auditable end to end
Digital participation systems now collect civic input at scale, yet they often fail to preserve inspectable links between contributions, the evidence behind them, the procedures that consider them, and the decisions that follow. When those links are lost, accountability and long-term institutional learning weaken, and so does public trust.
Linked Open Democracy addresses two coupled deficits: trust in digital participation infrastructures, and sustained participation over time. In Europe the need is timely, as Regulation (EU) 2024/1183 and democratic-resilience initiatives push toward trustworthy, interoperable, and privacy-preserving public digital infrastructure.
The project investigates whether a provenance-aware semantic infrastructure, built on proportional identity assurance, portable communities, grounded AI, and simulation-based validation, can make civic participation more auditable, explainable, and answerable. It is framed as a question to be tested, not a promise that technology alone will restore democratic trust.
The title names the project's intention. Linked Open Democracy plays on Linked Open Data, from Tim Berners-Lee's Semantic Web, asking whether participation, like data, gains value when it is open, linked, and machine-interpretable, so the aim is to empower participation rather than merely digitise it. The Digital Contract echoes Rousseau's Social Contract, recast for an age where some of democracy's chains are digital, pointing to a more explicit and inspectable relation between citizens, digital infrastructures, and institutions.
Working hypotheses
Four assumptions guide the design and are tested cumulatively through the planned artefacts and evaluation, rather than treated as settled conclusions.
Proportional identity improves legitimacy
Modelling eligibility, credentials, consent, revocation, and assurance as privacy-aware semantic relations can strengthen perceived legitimacy without requiring excessive disclosure of personal data.
KG-grounded AI improves deployment readiness
AI that works over curated knowledge graphs and returns evidence-linked outputs can be evaluated more systematically than answers drawn from unstructured context alone.
Inspectable feedback loops improve agency
When citizens can see how contributions, arguments, evidence, and institutional responses connect, they can better understand and act on their role in the democratic process.
Four cumulative outputs
One stack, built in layers
Each artefact builds on the one before it, from identity and trust up to simulation and validation.
Linked Open Identification
The trust baseline: privacy-preserving identity, credential provenance, assurance semantics, and revocation and consent control.
Prototype in developmentLinked Open Communities
Communities, participation events, governance constraints, and interaction histories modelled as linked semantic structures.
PlannedLinked Open Democracy
Participation intelligence: a civic app that empowers citizens to propose ideas, debate, bring evidence, take part in consultations, follow legislative links, and see institutional responses, helping people feel engaged with, and part of, the democratic process through technology.
PlannedLinked Open Simulation
Simulation to test democratic scenarios and compare policy pathways before broader institutional adoption.
PlannedWhat we are building toward
A manifesto for an auditable digital public sphere
The principles
Linked Open Democracy is guided by a simple conviction: digital participation earns public trust only when it is transparent, auditable, and answerable to the people it serves. These principles express what a trustworthy digital public sphere should make possible. They are commitments to investigate and build toward, not claims already solved. Each is tagged to the layer of the project that pursues it.
Identity & personhood
- A person should be able to prove they are a real human being online without having to reveal who they are.
- A citizen should be able to tell a real individual apart from an automated agent, and to know which one they are speaking with.
- Identity should be asked for in proportion to what is at stake, never demanding more disclosure than an interaction genuinely requires.
- Every credential a person presents should carry its own provenance, so its origin, assurance, and validity can be inspected and, when necessary, revoked.
- Consent to share a personal attribute should be explicit, narrowly scoped, revocable, and recorded: never assumed, and never permanent.
Trust & accountability
- Trust online should be earned and open to inspection, not decreed by whoever happens to own the platform.
- No score, label, or automated judgement about a person should be a black box; it must be explainable, logged, contestable, and open to appeal.
- Being verified as a real person is not the same as being truthful or acting in good faith; infrastructure should support human judgement, not replace it.
Communities & expression
- People should be free to speak and to associate, and to build their own communities to debate ideas without asking a private platform for permission.
- The spaces where public debate happens should be portable, transparent, and governed by their members, not locked inside a single company.
- A community should be able to set and publish its own rules of participation, and be held to them in the open.
- The history of a debate (who took part, under what rules, and how positions changed) should remain legible over time.
Democracy & institutions
- Technology should help people feel part of the democratic process, turning distant political decisions into something they can follow, question, and help shape.
- Citizens should be able to propose ideas, debate them, and bring evidence, and to see how their input links to consultations, legislation, and the decisions that result.
- Institutions that invite participation owe a visible and traceable response, so contributions are genuinely considered, not merely collected.
- AI should augment civic participation rather than stand in for it, staying grounded in real evidence, citing its sources, and leaving the final judgement with people.
Validation & humility
- Democratic infrastructure should be tested before it is trusted; policy pathways and failure modes deserve examination in simulation before they touch real institutions.
- No layer of this should be adopted on faith; each must be evaluated both technically and with the people it affects, and revised when the evidence demands it.
Outputs & links
What is available today
Live instruments now, with software, documentation, and publications added as the project advances.
Live
Survey 1: Digital identity & trust
A study on digital identity, democratic trust, civic participation, and perceived online manipulation. Participants only choose among simulated sign-in options; no real credentials are entered.
Open the survey →
PDF
Project poster (2026)
A one-page overview of the research problem, the LOI/LOC/LOD/LOS stack, the methodological design, and the evaluation plan.
View the poster →
Coming soon
LOI software & docs
The Linked Open Identification proof of concept: a standards-based, privacy-preserving login and provenance profile, with developer documentation.
In development
Planned
LOC: Communities layer
Portable, member-governed civic spaces modelled as linked data, with transparent rules and legible debate histories.
Planned
Planned
LOD: Democracy layer
Policy and participation ontologies, civic text-to-graph transformation, and evidence-grounded AI assistance for institutional decisions.
Planned
Planned
LOS: Simulation layer
Scenario testing and policy-pathway comparison to stress-test the integrated architecture before broader adoption.
PlannedPublications & events will be listed here as the project produces them. The first paper, on Linked Open Identification, is in preparation.
People
Who is behind the project
Paulo Jorge Pereira Martins
PhD candidate · Author
Doctoral researcher in Informatics Engineering at the University of Coimbra, with experience in Natural Language Processing, Semantic Web, Knowledge Graphs, Health Informatics, and Digital Democracy for civic participation. Paulo is a work-student, researching and implementing the Linked Open Democracy stack while working at Salsify.
Highlights
- Senior Software Engineer at Salsify (GDSN) and Viz.ai (Healthcare and AI).
- Assistant Professor and R&D collaborator at the University of Minho (DI and CEHUM).
- R&D Bosch Project: tech founder of RideCare.
- Founder of Retro Tales.
- Led the technical inception of BragaTV.
- Arquivo.pt 2021 project award (Major Minors) and Prémio Matilde Rosa Araújo 2025 (short story).
Hugo Gonçalo Oliveira
Professor · Supervisor
Associate Professor at the Department of Informatics Engineering of the University of Coimbra and researcher at CISUC, where his main research areas are Natural Language Processing and the Semantic Web. Hugo supervises several MSc and PhD students and guides the project's scientific direction.
Highlights
- Coordinator of the MSc in Artificial Intelligence at the University of Coimbra.
- Author of 230+ peer-reviewed papers in NLP and the Semantic Web.
- Executive editor of the journal Linguamática, and a regular member of major NLP program committees.
- Best PhD thesis in the Computational Processing of Portuguese (2014).
- Special Recognition in Education by the World Cultural Council (2022).
Follow the project
Pre-register your interest in upcoming studies and join the mailing list to follow the project's progress. For research enquiries, reach out by email.