Author: Karl Grossner

Contributing in Linked Places format

HGIS de lasIndias spatial footprint
Fig. 1 – HGIS de las Indias spatial footprint

We are pleased to announce version 1 of the Linked Places (LP) format, to be used for contributions to World-Historical Gazetteer (WHG) and to the Pelagios “registry index” that underlies its Recogito annotation tool and Peripleo search interface. We were joined in this effort by Rainer Simon of Pelagios and Graham Klyne and Arno Bosse of Oxford’s Cultures of Knowledge, so a big shout of thanks to all three.

A basic specification of the format has been published to GitHub, along with some relevant files:

The spatial footprint of the HGIS de las Indias contribution appears in Figure 1; more about that data and ongoing efforts to make it the “seed” for a virtual LatAm Gazetteer within WHG are forthcoming soon.

An Interconnection Format, Not “The One” Model

Historical research projects producing gazetteer data have distinctive data models reflecting their source data and project-specific requirements. We are completely agnostic as to contributors’ internal models and formats. The Linked Places format provides a uniform way to build links between different gazetteers.

Temporal Scoping

WHG and Pelagios are aggregating contributions which have great variety in scope and granularity. With LP format, we are striking a balance between accommodating detail and offering simplicity. Two underlying conceptual models: of Place and of attestations. We say (hopefully not controversially) that places have, over time,

  • one or more names in various languages
  • one or more functional types
  • zero or more known spatial footprints (geometries)
  • one or more relations to other places

Gazetteer data developers gather attestations of these properties, often temporally scoped. That is, names, types, geometries, and relations were/are true for some timespan—whether we have that detail to hand or not. Thing is, some projects do and some don’t, and we must accommodate both cases.

The simple subject, predicate, object form of RDF is not adequate for these kinds of relations and so we have reified them as attestations: NameAttestation, TypeAttestation, Setting, and RelAttestation. Each can be temporally scoped with a “when” element and include citation information, although this is not required. We also permit (and encourage) a global “when” element for the record—effectively the union of temporal attributes for names, types, geometries, and relations.

Links

We ask that all contributions include, if at all possible, at least one closeMatch or exactMatch relation to a published place record having a de-referenceable URI. These links are “coin of the realm” so to speak. If an incoming record shares such a link with an existing record in our registry index, we  add it to the existing record. If it shares no such link with any existing records and doesn’t otherwise match name, type, and geometry, a new registry index record will be created using it as a seed.

Three other kinds of LinkAttestations to existing web resources are recognized: subjectOf, primaryTopicOf, and seeAlso.

Record-Level Properties

Several “record-level” properties round out this format, as detailed in the GitHub specification: title, ccode (modern country), description(s), and depiction(s).

Moving Forward

Please get in touch  if you are interested in contributing to WHG. The Linked Places format is not set in stone; this v1 is subject to revision based on our experiences ingesting contributions. The ingest of HGIS de las Indias data as a gazetteer offered to users of Recogito has succeeded, following significant manual effort to make the transformation from that project’s complex model to LP format. Temporally scoped namings and parthood relations in that dataset made it over, but are not reflected in Recogito. Simpler example datasets will be added soon. The WHG web interface and API are being designed to expose all contributed place attributes, but they won’t be available for several months.

Linked Places Annotations

Next up in our format-creation effort is an update to the standard format for contributions of annotations. Just as Peripleo displays coins and inscriptions associated with places in its registry index, the WHG graphical search will such annotation records, but in our case for historical routes, datasets, and bibliographic records. The format previously used by Pelagios/Peripleo will be updated soon—details to follow.

Contributing to World-Historical Gazetteer: a Preview

The World-Historical Gazetteer project (WHG) will soon begin aggregating and indexing historical gazetteer datasets, and exposing them as Linked Open Data via graphical and programmatic web interfaces — just as Pelagios Commons’ Peripleo project has done for a few years. And like Peripleo, WHG will also index contributions of annotation records that associate historical “items” with place identifiers. Typical items for Peripleo have included coins, coin hoards, and inscriptions of the Classical Mediterranean. Items records WHG will focus on include journey events, regions, and datasets. In fact, annotated items could be anything for which location is relevant, e.g. people and various types of events.

We are almost ready to begin accepting contributions; this post previews the pipeline and formats involved.

Contributions to WHG can include, in some combination: 1) gazetteer data, i.e. place records drawn from historical sources; 2) annotation records that associate a published record about an item with a place identifier; 3) collections of item metadata records referenced in annotations; and 4) a file describing the contributed dataset(s) in Vocabulary of Interlinked Datasets” format (VoID).

Over the past several weeks we have collaboratively developed a new Linked Places format (LPF here for short) with Rainer Simon of Pelagios, to be used for contributions of historical place data to both WHG and Pelagios’ Peripleo. The Linked Places format is designed around the JSON-LD syntax of RDF (it is also valid GeoJSON, with temporal extensions, as explained in the GitHub README). The new format makes use of several existing vocabularies and also introduces some terms specific to our shared purposes.

Several expert colleagues contributed valuable input, including Graham Klyne, Richard Light, Lex Berman, Arno Bosse, and Rob Sanderson [1]. We are in the process of updating the template Peripleo has used for annotation contributions (formerly Open Annotation in RDF Turtle, now its next-generation W3C Web Annotation in JSON-LD). Both are discussed in a little more detail below.

Contributing historical place records

There will be two separate workflows for contributions: from larger projects and from smaller ones. The distinction is whether a project has the capability and resources to meet two criteria which are accepted norms for publishing Linked Data: 1) publishing data in some syntax of RDF (in our case the new LPF); and 2) providing a unique URI and associated “landing page” for each resource described.

Case 1: Larger projects

If your project has (or will have) a web presence that provides public pages describing your individual places and/or “items”, (routes, regions, etc.), then we ask that you perform a transformation and export of your data in the standard formats mentioned above – Linked Places, a future annotation format (see Contributing Annotations below), and VoIF. Upon validation, we will ingest those records, link them with those already in the system, and expose them in a nice GUI and API. Details of WHG interfaces are forthcoming soon.

Case 2: Smaller projects

If your project does not entail creating a web site providing per-record landing pages, then we can accept your data contribution as CSV, mint unique URIs, and provide very basic landing pages for places and other items. The records will also be made available as JSON-LD (bonafide RDF) via our API. We will provide a Python program for converting CSV to LPF, but note that the CSV will have to conform to a template that aligns with LPF (available soon). Conversion from your native format to our CSV template will probably be more manageable than to LPF. In other words, upon submitting CSV data we can parse, a semi-automated conversion and ingest procedure will result their publication as Linked Open Data.

Contributing annotations

WHG will index metadata describing historical “items” annotated with gazetteer record identifiers. These annotation records assert, in effect: “this item is/was associated with this place, in this way;” and optionally, “at this time.”

The result of such annotations can be seen in the current Peripleo interface, where upon navigating to a given place, you can view metadata (including images) for coins and inscriptions associated with it in e.g. a foundAt or hasLocation relation. Annotations exposed in the WHG web interface will include historical journeys for which the given place was a waypoint, and regions, works, and datasets including or referring to the place.

Annotation contributions will comprise two sets of data: 1) collections of brief Item metadata records; and 2) collections of annotation records in W3C Web Annotation format. The contribution template in use by Pelagios’ Peripleo now is currently being updated to better account for typing of items and relations. Details of that new Linked Places annotation format (LPAF?) will be published soon. Collaborators in that modeling effort are most welcome!

[1] Twitter handles, in order: @gklyne, @RichardOfSussex, @mlex, @kintopp, and @azaroth42