Semantification at the creation points: which tool, run by whom, creates and checks paper ids, DOIs and JSON-LD

[WolfgangFahl]

✍️🤖 Agent/Sotsog: per Wolfgang Fahl

Question

Semantified metadata - paper ids, landing pages, DOIs, the JSON-LD of a
volume - has to be created where the volume is created, not recovered
afterwards from the published HTML. At each creation point a different
person runs a different tool. Where does semantification have to plug in,
and into whose tool?

Creation points

#	creation point	who	tool today	semantified data created or checked there	hop on CEUR-WS Graph Walks
1	volume prepared: index.html, paper PDFs, agreements	workshop organizers	ceurtools on their own machine (row 0 of the editor agreement asks whether they ran them; Linux only, OrbStack/WSL)	paper id as li id of the TOC entry, PDF file name, CEUR span marks, order of the papers; check-index-errors compares li id and PDF file name	P1 Proceedings draft →contains Paper →author Scholar →signed AuthorAgreement (none)
2	submission checked	CEUR editor	ceurtools on the CEUR-WS server	the same checks again; check-pdf-errors tests whether a paper already carries a DOI in its footer	C1 Ticket (none) →for Proceedings →contains Paper
3	volume published	CEUR editor	publishing step	URN; JSON-LD of the volume, #19	C2 Proceedings →commit (none) volume page, URN (none)
4	DOIs registered	DOI registration	register-dois.py, #14, input index.jsonld: volume DOI prefix/ceurws_NNNNN, paper DOI prefix/ceurws_NNNNN_PPP with PPP the running number of the papers with authors in JSON-LD order (8f3f529), landing page https://ceur-ws.org/Vol-N/<paper id>/, CEURDOI inserted at the li with that id	DOI and landing page per paper	C2 Paper →P356 DOI, landing page (none)
5	archive semantified	technical editors	volume page mapping (#28), ceur-spt	everything above for the published volumes, with the faults of #12, #22, #35 to #38	T2, and the broken hops of the concrete walks W1 to W5

Why it depends on who runs the tool

• register-dois.py reads index.jsonld; the order of its paper list fixes
  the paper DOIs and the paper id fixes the landing page; both are decided
  at point 1 by the organizers, so a wrong order, a missing or wrong li id
  or a paper published elsewhere in the list (volume page mapping: entries of papers published elsewhere are taken as papers of the volume #35) becomes a registered
  DOI or landing page at point 4 unless it is checked before
• a rule enforced at point 1 needs a tool the organizers can run on their
  machine; the same rule at point 2 or 5 can stay a Linux script on the
  server
• the archive at point 5 shows what happens without a check at point 1:
  volume page mapping: paper entries without li id are dropped #36 entries without li id, volume page mapping: volumes without CEUR marks give no papers #37 volumes without CEUR marks, volume page mapping: entries of papers published elsewhere are taken as papers of the volume #35 papers
  published elsewhere taken as papers of the volume

Walks

The creation points are hops of the stakeholder walks on
CEUR-WS Graph Walks
(walks P1, C1, C2 of the proceedings editor and the CEUR editor). In the
Wikidata version of these walks the workflow hops - AuthorAgreement,
Ticket, volume page commit, landing page, URN - have no property; they
exist only where the tool at that creation point records them. The
concrete walks W1 to W5 on the same page show where published data
already breaks, e.g. proceedings without short name and without volume
url (W3).

Proposal

• one rule list with a criterion per rule and the creation point and the
  person it applies to; ceurtools, the checklist of Checklist for the 0.0.1 MVP prototype: rules for JSON-LD, generated page and year page (basis for #13) #23 and the volume
  page mapping implement it
• the rules a DOI depends on - paper id, paper order, no prior DOI -
  are checked at point 1, not first at point 4
• the workflow states of State diagram for the new volume workflow: JSON-LD first or HTML first (basis for #19) #21 and the ticket alignment of Align Technial and Editorial Tickets - OsTicket #1118 of 2023-04-14 #9 and technical workflow on volume submission #10 get
  the creation points as their states

Open questions

• which rules move to point 1, i.e. become mandatory for organizers
• which tool creates the index.jsonld that register-dois.py reads, and at
  which point
• whether the running number stays stable when a paper is added or
  removed before registration