SciCom Wiki: Collaborative Science Communication

• SciCom Wiki is a collaborative, semistructured digital platform that integrates linked data, open annotation, and expert review to support science communication.
• The system integrates a dual-layer architecture combining a linked-data wiki store and full-text repository for granular metadata curation and federated search.
• It operationalizes FAIR principles through persistent identifiers, open APIs, and semantic interoperability to ensure data findability, accessibility, and reuse.

A SciCom Wiki is a collaborative and semi-structured digital knowledge platform designed to support the collective creation, curation, and dissemination of scientific information, with particular emphasis on representational interoperability, stakeholder-driven workflows, and mechanisms for integrating expert review. The concept encompasses both domain-neutral and domain-specific systems that prioritize open annotation, metadata enrichment, and semantic interoperability. SciCom Wikis are especially relevant for documenting and verifying emerging forms of science communication, including videos, podcasts, datasets, and computational code, and provide infrastructural backbone for the Science Communication Knowledge Infrastructure (SciCom KI).

1. Architectural and Functional Foundations

A prototypical SciCom Wiki combines a collaborative wiki platform with an underlying semantic or linked-data layer for structured knowledge representation. Notable architectural elements include:

• Linked Data Wiki Store (e.g., Wikibase): Central graph database maintains metadata and semantic annotations for scientific artifacts, encompassing more than 200 distinct properties for media items (e.g., hasTranscript, license, contributors, topics) (Wittenborg et al., 12 Nov 2025).
• Full Text Wiki: Stores curated, copyright-cleared transcripts and long-form text, supporting granular annotation (e.g., paragraph-level) and enabling full-text search and advanced textual semantics.
• Dashboard and Microservices: Expose API-driven interfaces for search, bulk import/export, federated result merging, and downstream integration with external services.
• User and Stakeholder Model: Explicit roles (viewer, researcher, teacher, content creator, curator, developer) are mapped to platform permissions and functionalities (see Table 1).

Table 1: Stakeholder Role-Activity-Service Mapping (Wittenborg et al., 12 May 2025)

Role	Primary Activities	Platform Services
Viewer	Search, Filter, Consume	Dashboard (read only)
Researcher	Explore, Annotate, Compare Fact-check	Wikibase Query, Full Text Wiki
Content Creator	Upload, Describe, Source-link	Aggregation Service (Import)
Curator	Structure Ontology, Define Properties	Wikibase Admin UI
Developer	Extend APIs, Plugins	Microservice, Dockerfiles

Functional requirements (FR) include: rich metadata search (FR1), transcript search (FR2), filtering (FR3), inter-media comparison (FR4), annotation of claims and sources (FR5), integration with fact-checkers (FR6), bulk metadata import/export (FR7, FR8), trust metrics (FR9), and linkage to supplementary materials (FR10) (Wittenborg et al., 12 Nov 2025).

2. Semantic and Social Layer: Verification, Annotation, and Provenance

SciCom Wikis achieve scalable, reliable knowledge structuring via hybrid social-semantic workflows:

• Semantic Annotation Workflow: Uses lightweight forms or WYSIWYG/AJAX editors to annotate content with ontology-aligned metadata (Leclercq et al., 2010). Semantic triples are generated and validated against domain ontologies (typically in RDF/OWL), facilitating automated reasoning, enforcement of integrity constraints, and guided navigation (e.g., faceted browsing, query panels).
• Permission Model: Strongly tiered, with roles encompassing unregistered users, registered contributors, verified experts, and editorial council; original author control over primary data is maintained via namespace and per-tag permissions in platforms such as Fluidinfo (Seidel, 2012).
• Versioning and Forking: Encourages continuous refinement with clear separation of stable, citable versions (e.g., "clusters" in Citizendium) from draft or developmental forks, achieved via explicit forking models:

$$\text{Draft}_0 \;\xrightarrow[\text{ExpertReview}]{\text{approve}} \bigl(\text{Cluster}_\mathrm{stable},\,\text{Draft}_1\bigr)$$

(Morris et al., 2010)

• Provenance and Annotation: Every atomic change or annotation is attributed, timestamped, and tracked under contributor-controlled namespaces or subpages (Seidel, 2012, Morris et al., 2010).

3. FAIR Principles and Knowledge Representation for New Media

Modern SciCom Wikis operationalize the FAIR principles as follows:

• Findable: Unique persistent identifiers (e.g., URIs for Wikibase items, transcripts) and SPARQL endpoints (Wittenborg et al., 12 Nov 2025).
• Accessible: Open APIs, absence of paywalls, compliance with content negotiation protocols (e.g., JSON-LD, RDF/XML).
• Interoperable: Data models rely on RDF and OWL, leveraging existing vocabularies and property/class patterns from Wikidata or domain ontologies, and support cross-linkage to DOIs, ORCIDs, and external knowledge graphs.
• Reusable: Licenses are explicitly declared (CC0/CC-BY), and provenance (curator, timestamp) is machine-readable and queryable; annotation separation preserves data integrity and supports multi-user, concurrent annotation (Wittenborg et al., 12 Nov 2025, Seidel, 2012).

FAIR compliance can be numerically expressed as:

$$\text{FAIR\_Score} = \frac{1}{4} \sum_{i=1}^{4} C_i$$

where $C_i$ corresponds to core criteria for unique IDs, rich metadata, open licensing, and machine-readable formats.

4. Fact-Checking and Computational Verification Workflows

SciCom Wikis increasingly integrate computational fact-checking, emphasizing:

• Media-to-Knowledge Graph Conversion: Extracts entity-relation triples from source media (e.g., transcripts) using LLM-driven NER and minimally guided ontologies, formalized as:

$$G = (V, E),\quad V = \{\text{entities, classes, literals}\},\ E \subset V \times P \times V$$

(Wittenborg et al., 12 May 2025)

• Neurosymbolic Fact-Checking: Compares extracted, untrusted knowledge graphs ($G_U$) with ground-truth graphs ($G_T$; e.g., derived from trusted scientific reports like IPCC AR6). Veracity for triple $(s,p,o)$ is scored as:

$$S_{ver}(s,p,o) = \begin{cases} 1, & (s,p,o) \in G_T \ e^{-\alpha d_{G_T}(s,o)}, & \text{otherwise} \end{cases}$$

with $d_{G_T}$ shortest-path in $G_T$, and $\alpha>0$ a scaling parameter. Scores are annotated back to the Wikibase (Wittenborg et al., 12 May 2025).

• Evaluation and Provenance: Expert/lay user studies confirm utility and acceptance; accuracy for exact-match fact-checks reaches 0.81 precision, area under curve (AUC) ~0.78 for path-based inferences.

5. Interoperability, Integration, and Community Practices

SciCom Wikis are explicitly federated and extensible by architectural and social design:

• Extensibility: Modular microservice and API architectures support plugin development, federated querying, and continuous integration with domain repositories (e.g., TIB AV-Portal, World Lecture Project, ADS Labs for astronomy) (Wittenborg et al., 12 Nov 2025, Henneken et al., 2012).
• Community Curation: Platforms accommodate course-based collaborations, crowdsourced metrics, and fine-grained rating or trust frameworks (e.g., Web of Trust analogs). Annotation workflows are scaffolded to allow novice and expert contributors, with progressive disclosure of semantic features as needed (Morris et al., 2010, Leclercq et al., 2010).
• Living Ecosystem: Automatic ingestion, federated SPARQL/REST endpoints, and clear UI abstractions balance the needs of high-volume data, non-technical contributors, and advanced research users.

6. Evaluation, Usability, and Impact Metrics

Rigorous usability and evaluation studies are embedded into SciCom Wiki deployment and roadmap:

• Evaluation Protocols: Structured task-based experiments (search/filter/use) assess objective (task time, completion rate) and subjective (user experience, satisfaction) metrics. Notably, four of five benchmark tasks are completed under a three-minute threshold, with mean user experience benchmark (UEQ) at 1.78 ("excellent" on attractiveness, efficiency, and dependability) (Wittenborg et al., 12 Nov 2025).
• Scalability and Performance: Architectural decisions (e.g., federated microservices, efficient data models, and advanced caching strategies) are tailored to operate at millions-of-item scale (Wittenborg et al., 12 Nov 2025, Seidel, 2012).
• Criteria Satisfaction: User studies establish demand for enhanced correctness, complexity, and source transparency in metadata; legal and ethical frameworks are identified as essential for full implementation of advanced features.

7. Best Practices and Future Directions

Sustained development of SciCom Wikis is guided by several best practices:

1. Implement real-name registration and expertise validation for governance and quality assurance (Morris et al., 2010).
2. Layer lightweight, expert-driven review to ensure accuracy, neutrality, and accessibility.
3. Enable citable, stable cluster forking while supporting ongoing, collaborative draft work.
4. Embed structured subpages, semantic links, and modular annotation layers for context-rich navigation (Morris et al., 2010, Leclercq et al., 2010).
5. Codify transparent, chartered governance structures with clear dispute and policy mechanisms.
6. Systematically expand metadata models and automation pipelines to support emerging and underrepresented non-textual scientific media (videos, podcasts).
7. Promote open-source, community stewardship for long-term sustainability (Wittenborg et al., 12 Nov 2025).

By integrating modular semantic infrastructure, transparent collaborative processes, and end-to-end FAIR compliance, the SciCom Wiki paradigm operationalizes a scalable, extensible, and verifiable backbone for science communication in the era of proliferating digital media and data modalities.