# Publication Interface ## Purpose The Publication Interface connects Coulomb-native research to conventional scientific literature and institutions. The governing rule is: > The living research graph is canonical. Papers are import/export containers. This allows Coulomb to interoperate with traditional science without making the paper the primary internal unit of knowledge. ## 1. Inbound: Paper Digestion An external paper should be treated as an externally authored artifact that can be decomposed into reusable research objects. ### 1.1 Ingestion pipeline ```text external paper ↓ identity resolution ↓ metadata extraction ↓ content parsing ↓ claim / method / evidence extraction ↓ relation mapping ↓ provenance capture ↓ Coulomb research graph ``` ### 1.2 Candidate input formats - DOI - arXiv identifier - URL - PDF - BibTeX - RIS - JATS XML - Crossref metadata - DataCite metadata - repository metadata ### 1.3 PaperDigest model A useful normalized representation: ```yaml source: doi: arxiv: url: identity: title: authors: institutions: publication_date: venue: digest: research_question: propositions: methods: evidence: limitations: conclusions: relations: depends_on: supports: challenges: extends: uses_dataset: uses_software: provenance: digest_actor: digest_model: created_at: confidence: human_verified: ``` ### 1.4 Epistemic separation The system must preserve the distinction between: - what the paper states, - what an extraction agent inferred, - what external evidence supports, - what Coulomb participants currently accept. No ingestion pipeline should silently convert extraction into endorsement. ## 2. Recursive Literature Digestion Once ingested, a paper may trigger additional work: ```text references ↓ missing foundational work cited-by graph ↓ later extensions and criticism extracted propositions ↓ matching Coulomb propositions methods ↓ replication candidates limitations ↓ open problems ``` Traditional scientific publishing becomes an external sensor network for Coulomb. ## 3. Outbound: Paper Projection A paper is generated from a selected region of the research graph. ```text research graph ↓ scope selection ↓ narrative synthesis ↓ paper representation ↓ target-specific rendering ``` The exported paper may include: - research question, - context, - propositions, - methods, - evidence, - results, - criticism, - limitations, - contribution statements, - references, - supplementary artifacts. ## 4. Publication Stages Publication state is orthogonal to epistemic state. A paper may be epistemically weak but externally published, or epistemically strong but unpublished. ### P1 — Internal Circulation Purpose: - collaborator review, - narrative coherence, - limitation discovery, - synthesis of current state. Characteristics: - restricted visibility, - mutable revisions, - stable Coulomb version identifiers, - no expectation of venue compliance. Example: ```yaml publication_stage: P1 publication_state: internal-circulation visibility: collaborators ``` ### P2 — External Preprint Purpose: - public disclosure, - wider criticism, - prior-art visibility, - citable public version. Characteristics: - public artifact, - external identifier where appropriate, - explicit versioning, - linkage back to Coulomb provenance. Example: ```yaml publication_stage: P2 publication_state: external-preprint external_identifiers: arxiv: doi: ``` ### P3 — Venue Candidate Purpose: - target a specific conference, journal, workshop, or other conventional venue. Characteristics: - venue-specific format, - scope matching, - submission constraints, - anonymization if required, - targeted revision cycle. Example: ```yaml publication_stage: P3 publication_state: venue-candidate target: venue: track: deadline: ``` ### External Version of Record A published conference/journal version should be tracked as an external event, not as a superior Coulomb epistemic state. ```yaml external_publication: venue: doi: accepted_at: published_at: ``` ## 5. Publication Adapter Conceptual interface: ```text COULOMB GRAPH │ Publication Adapter ┌───────┴────────┐ ▼ ▼ IMPORT EXPORT │ │ DOI / PDF / ... graph selection │ │ ▼ ▼ Digest renderer │ ▼ venue-compatible paper ``` ## 6. Export Targets Potential exporters: - Markdown - LaTeX - PDF - DOCX - JATS - arXiv bundle - BibTeX - CSL JSON - venue-specific templates The long-term interface may resemble: ```text publish --stage P2 --target arxiv publish --stage P3 --target neurips ``` without making those targets part of the core ontology. ## 7. Scientific Identity Interoperability Where useful, map Coulomb identities and contributions to: - ORCID for researchers, - DOI for persistent outputs, - ROR for institutions, - CRediT for conventional contribution statements. Coulomb's internal contribution graph may remain richer than any external standard. ## 8. Agent Authorship and Contribution Coulomb should record machine contributions explicitly. External publication adapters must comply with target-venue rules while preserving internal provenance. Example internal provenance: ```yaml contributors: - actor: human:alice contributions: [conceptualization, validation] - actor: agent:critic-v4 contributions: [literature-analysis, counterexample-search] ``` If an external venue does not recognize agent authorship, the adapter should transform presentation without erasing the internal record. ## 9. Design Principle > Interoperate at the boundary; do not duplicate legacy publication infrastructure internally.