models

Linked data model for EMPlaces

This directory contains diagrams and documentation and sample data for a proposed linked data model for EMPlaces.

See also links for related information.

NOTE: this is WORK-IN-PROGRESS, and NOT FINAL.

(In due course, the documentation itself may be moved into this page)

TODO

• Finalize some URIs to be used

• Finalize prefixes

• Finish assembling example data for Opole

  • Fix structure and vocabulary for bibliography (using BIBO)
  • Historical hierarchies
  • Fix vocabularies for timespans (PeriodO?)
  • Uncertainties, approximations, etc
  • Related resource, include optional license (see LPiF proposal)
    • also, type of resource?

• Pin down location/timespan vocabularies (the structure here follows ideas from Topotime/GeoJSON-LDT (also a work-in-progress?) but the vocabularies haven't yet been checked.)

• Review format for timespan data; align with emerging LPiF activity (loosely aliogned, but not same vocabulary for now)

• Update Data model notes; resolve remaining TODOs there, as far as possible.

• Generate data from EMLO data

  • first 500 done; awaiting feedback

• note 2 kinds of Julian calendars

  • currently using "Old style" and "New style", but this is probably not enough.
  • suggestion that we might add start-of-year date to calendar details, but even that isn't always enough: in some cases 25 Dec start of year occurs in the year BEFORE that indicated?
  • currently handled in Annalist definitions. Can revisit later if needed.

• Review handling of date uncertainty/aprroximation, and how it relates to PGiF proposal to use ISO 8601-2 (https://github.com/LinkedPasts/lpif)

  • currently using LPiF-related structures, rather than ISO date formats

• Type URIs for places that don't have same kind of info (cf. "related places", 20180726-St-Adalbert-example.ttl)

• Record metadata design (creator, contributor, license, etc.)

  • waiting to see what Timbuctoo provides

• Record time periods as identified resources (rather than inline blank nodes)

  • Handled in Annalist data; not yet in Opole example

• URIs for language codes (consider lexvo?)

  • holding pattern in place, and Annalist data. Revisit later.

• ...

• Decide how to flag "core data" in structures used for both core and additional data (needed for refresh of core data from source). It seems what is really needed is a reference source indication.

  • A declared set of properties of the central em:Place resource are always core data: see em:Place class declaration.
  • Qualified relations, settings and annotation resources have (optional) em:source properties: the corresponding values are considered to be core data if the em:source/em:link value is the same as the place's em:coreDataRef/em:link value.
  • Note that the place's em:coreDataRef/em:link and corresponding em:source/em:link values refer to the gazeteer defining document, not the gazetteer place Id.

• Decide on structure for place categories and annotation types (using skos:Concepts)

• Check that proposed Web Annotation extensions don't break anything (email with Robert Sanderson)

• Update diagrams

• Crosswalk between UI and data model

• Dealing with uncertainty

  • See example, em:competence
  • How to represent?
    • Confidence flag (high/medium/low)? Not really going to work for us?
    • Permutation of (Uncertain, Inferred, Approximate) per EMLO. But be explicit in all cases.
  • Where to represent: calendars, date of hierarchical relations
  • For places: Uncertain, Inferred
  • For dates: Uncertain, Inferred, Approximate (implicit the timespan value)

• Characterize calendars (inherited calendars; need to materialize for indexing; need source indication that it is inherited)

  • Handled by em:competence
  • general issue here about materialization of inferred/implied/deduced information.
  • note that name attestations are additions to the primary info; no such for calendars.

• Review the way that core data is represented

Notes (questions and ideas)

@@TODO: allocate URIs for attestations (where for our purposes, the oa:Annotation denotes the attestation)?

Priorities (2018-08-01, 2018-08-14)

1. Modelling of core data to deal with multiple core data sources
   • update diagrams
   • update sample data
2. Need to start getting data into RDF and/or Timbuctoo
   • e.g. 4 more towns like Opole to comparable detail
   • Create Annalist definitions for data entry
3. Bulk data into RDF/Timbuctoo
   • e.g. 2000 places core data from GeoNames
   • update core data extractor to generate multi-source structure
   • Wikidata cross-referencing for alternate authorities
4. Loose ends
   • Modelling and Opole data
   • Related places
   • Historical place record: minimum viable information
     • differs from current place
     • discuss with Arno
5. Generate EMPlaces data mapped from Annalist
   • Basic mapping seems to work
   • Figure how to handle local URIs (which are roughly blank nodes) [Note: depends on Annalist code updates in 0.5.13 in development branch]
6. Annalist data
   • Generate Annalist data from geonames; get full update data
   • Review Annalist definitions with Arno

• Sample bibliographies to enter
• Dealing with sources (see issues #19, #13, #7 ?)
• Integration of date conversions

Actions