Meeting notes 7 March 2022

[Karl Grossner]

Hello all,

This Linked Places working group held its first meeting today, attended by 12 people (!). Two sent regrets. 

Karl opened with some history of LPF and its relation to World Historical Gazetteer (WHG):

• developed principally by Karl, Rainer Simon, Richard Light, and Graham Klyne to supercede an earlier "Pelagios Gazetteer Interconnection Format" (PGIF). the use cases were WHG and Pelagios apps (Recogito, Peripleo) at the time.
• strategy was to take a widely adopted standard for geodata (GeoJSON) and extend it, principally by adding time
• this took the form of a "when" object, which could be added to an entire feature, to individual geometries (standalone or within a GeometryCollection), and to some other newly introduced element arrays: "names", "types", and "relations"
• WHG contributors' data is very typically includes information about objects, people, and events related to places.
• WHG is agnostic as to how contributors model, store and manage their data--asking only for an "extraction" of places from whatever they have.

The discussion that followed began with a consideration of Ryan Shaw's summary of LPF purposes, cf. https://github.com/LinkedPasts/linked-places-format/issues/43

One focus was on the various requirements of apps/platforms looking to implement LPF or some extended version of it. Specifically and currently:

• the British Library's, Locating a National Collection (Stephen Gadd, Gethin Rees, Rainer Simon)
• Periodo (Ryan Shaw; historical period gazetteer)

Also, research projects looking to implement or export LPF for WHG, e.g.:

• Trismegistos (Tom Gheldof)
• DAI (ChronOntology, e.g.; Wolfgang Schmidle)
• DigiKAR (Ingo Frank)
  
• Urban Gazetteer of Ghent (Vincent Ducatteeuw)

A major point at the heart of much discussion: historical researchers and cultural heritage collections are almost always concerned not only with places, but the things, people, and events located at them. NB. many WHG contributors are particularly gathering only geodata, e.g. Historical GIS projects.

Some points made, including from the chat, with some paraphrasing:

• Karl: Ryan's summary is consistent with the WHG conception of LPF as strictly concerned with place data.
• Gethin: Three software projects pursued by partners of the Pelagios Network have some common interests: Recogito (annotation), WHG (gazetteers), Peripleo (visualization of places and "traces")
• Karl: Is this a "union use case"?
• Gethin: I should add that Periodo/temporal visualisation in Peripleo/annotation in Recogito also has a lot of potential here
• Ryan: if the RDF representation of LPF can be stabilized, then we can work on mapping that representation to some of the other major vocabularies (CRM, schema.org, OGC, OWL Time, etc.) and significantly ease the process of translating place data, even automating it to some extent
• Vincent: Rainer, would you have some example objects (from the Locating a National Collection project) that show the difficulty in separating Linked Places and Linked Traces?
• Rainer: A typical example would be archaeological records, which all have their findspots, i.e. mapping to a gazetteer has only limited value (for some use cases, at least).
• https://github.com/britishlibrary/peripleo-lanc/blob/main/source_data/original/PAS/PAS_Warwickshire.lp.json
• Rainer: Here's a smaller coin dataset: https://raw.githubusercontent.com/britishlibrary/peripleo-lanc/main/source_data/original/SW_Coins-Monasteries/alexandrian-tetradrachms.lp.json
• Ryan: If you have arbitrary geo-located data that is not a gazetteer, do you need LPF? Doesn't plain old GeoJSON(-LD) give you everything you need?
• Rainer: @Ryan, yes, I guess that's true. It's still useful to reuse definitions about typing, time, names etc. So it is "syntactically" a gazetteer. But semantically it's not places.
• Ryan: “It's still useful to reuse definitions about typing, time, names etc.” So maybe one possible direction is to break up LPF into those patterns, enumerating and defining them, and tools could focus on implementing some subset of them, rather than everything having to support a monolithic "standard format"?
• Ingo: My major requirement is already there as an user story -- maybe to focus on in the next session: https://github.com/LinkedPasts/linked-places-format/issues/25
• Vincent: Projects potentially contributing gazetteer data to WHG hold data in many formats, including CIDOC-CRM. A 'standard' mapping of CIDOC place/time-related terms to LPF would be helpful. (cf. Issue #25)
• Ryan: @Ingo I think this is an important use case / requirement
• Richard: It would be useful to gather these and other examples of data from various projects, to better consider commonalities that could make it into a "core" <-- paraphrasing

An Issue #45 has been added to the GitHub repo, to gather links to and snippets of relevant datasets to be considered.

A next meeting wil be scheduled for a few weeks from now, doodle poll to follow.

If I've missed things, please add them in response to this message, thanks!