Linked Places: A Modeling Pattern and Software for Representing Historical Movement [email protected],UnitedStatesofAmericaMerrickLexBermanmberman@cga.harvard.eduHarvardUniversity,UnitedStatesofAmericaRainerSimonrainer.simon@ait.ac.atAustrianInstituteofTechnology,Austria
Introduction This paper reports onwork in progress aimed atfacilitatingthecreation,sharing,linking,andanalysisofdataaboutthemovementofpeople,ideas,culturalpractices,andcommoditiesbetweenplaces,over thecourseofhistory.ProductsoftheLinkedPlacesprojectinclude: conceptual and logicalmodels for historicalroutes;atemporalextensionofthepopularGeoJSONdata format, called GeoJSON-T; several variedexemplar data sets converted to GeoJSON-T format;prototypewebsoftwareforbrowsingandvisualizingthatdata;andPythonscriptstoconvertdatabetweenCSV,GeoJSON-T,andRDFcompatiblewiththePelagiosGazetteerInterconnectionFormat.Substantialinterimwork products are shared in the Linked Places andGeoJSON-T GitHub repositories and have beenreportedinsomedetailintwoblogposts(1,2).
Motivation A growing number of historical gazetteers are
being developed in the course of digital humanitiesresearchprojects(Berman,Mostern&Southall,2016).Theirspatialtemporalcoverageistypicallylimitedtoaparticularareaandperiodduetofactorsofscholarlyquality,cost,andrelevancetoagivenproject.Coverageextentsdovaryconsiderably, froma single city forafewgenerationstoaregionforseveralcenturies.Withfew exceptions, these gazetteers are unpublished as
such;insteadtheyarespatialtablescontainedwithin,andintegralto,thelargerprojectdatastore.
Becausehistoricalgazetteersaredifficultandtime-consuming to produce, it is vital they be published,whenpossible,inawaythatpermitslinkingthem—anactivity that the Pelagios project has made greatstrides in facilitating. An emergent network ofspecializedgazetteersholdsterrificpromise,notonlyforre-use,butultimatelyasadistributed,increasinglycomprehensive geographical (i.e. spatial-temporal)index to linked data from numerous domains,including history, archaeology, literary studies,philology,andseveralofthesocialsciences.Thefocusof such an index, and encyclopedic applications itenables,will be on individual places, typically at thescalesofcitiesandpointsofinterest.
Suchsystemsarehighlydesirable,butgivenalargevolume of data about individual places we can alsobeginharvesting,creating,andsharingdataabouttheconnectionsbetweenthem.Weshouldbeabletoaskofhistoricalgazetteers:What journeysandhistoricalroutes has a given place been a waypoint on? And,what flows of people, ideas, and commodities has itbeenasourceorsinkfor?
ButtheLinkedPlacesandGeoJSON-Tprojectshavebeen undertaken with an even larger, “moonshot”vision in mind: a system allowing scholars and thegeneralpublictovisualizeandanalyzetheemergence,growth and spread of human settlements, theirchanging attributes, and the dynamic connectionsbetweenthem,includingthediffusionoftechnologiesandculturalpractices.
Torealizetheseideas,weneeda)lotsofdata,andb)methodsandmeansformergingorlinkingthem.Insome respects, we are starting from scratch; dataabout historical movement is sparse and stored indisparateforms.Muchofitwillbenewlygenerated,forexample by parsing texts, transforming tabularrecords, or digitally tracing lines on historicalmaps.Merging and linking operationswill require that theform of data from different sources (or abbreviatedcataloguesthereof)beeitherstandardized(inthecaseof merging), or similar enough that automatedalignmentisfeasible.
The majority of works on geographic networksconcernsphysicalmedia likeroadsandrail,whereasmovement data is eventive. Geographers havemodeled migration flows and disease diffusion forseveraldecades,providingtheoreticalbases fortheiranalysis that are outside our present scope. Anoverviewofthatworkisfoundin(Lowe&Moryadas1975).Anexcellentandmorerecentworkonmobility
andgeographicmovementisTimCresswell’s“OntheMove”(2006).Wearenotawareofanyeffortstomodeldata for historical routes computationally, howeverthe coreabstractionwebuildupon is the traditionalgraph/networkmodelofnodesandedgescreditedto18thcenturyworkofEuler(Biggs,etal1986).
A Modeling Pattern Data modeling is as much an art as a science(Simsion&Witt2004),butsomecorebestpracticesare well-known. A typical first step is establishingwhat entities are to be represented, what theiressentialattributesare,andwhatrelationshipsobtainbetweenthem(cf.Chen,1976).Thisstepisoftenbestaccomplished collaboratively, in an iterative processundertaken by domain experts. Our results wereimmediately published to blog posts and relevantlistservs,andtheresultinginputwasusefulinrefiningthemodel. When the modeling context is an individualresearch project, it hardly matters what names aregiven thoseentities and relationships—only that thedata store’s internal logic be sound and wellunderstoodbyprojectmembers.Butif,asinthiscase,thesystemwillaccommodatedatafrommanysourcesor be accessed by others, we need to find broadagreementonaconceptualmodelandavocabularyforits constituents between as many prospectiveparticipants as possible—that is, to describe theontology of the research domain. Although muchontology engineering of this sort has involvedcomprehensive high-level ontologies such as theCIDOC-CRM,thedevelopmentandimplementationofsmallontologydesignpatterns(ODP)hasbeengainingfavorsincetheintroductionofthatparadigmbyAldoGangemi (2005). Such patterns, by any name, are“reusablesuccessfulsolutionstoarecurrentmodelingproblem” (definitionprovidedby theAssociation forOntologyDesign& Patterns (ODPA) ) which can beusedaloneorassembledinmodularfashionforlargerrequirements. Examples include patterns for “Place,”“Event,”“Participation,”and“Region.” And so the first step taken in the Linked Placesprojecthasbeentodevelopanontologydesignpatternforthehistoricalmovementofsomethingbetweentwoormoreplacesoversomephysicalchannel,eitherforsome time during or throughout a timespan. Thepattern,visualizedinFigure1,comprisesthefollowingconceptualunderstandings: A route describes an attestation of one ormoreoccurrences of the movement of something (e.g.people, commodities, information) between two or
more places, either for some time during orthroughoutatime_period.Routesarecomposedofoneormore segment, eachofwhich is composedof twoplacesandapath(correspondingtonodesandedgesin network parlance), the locations and temporalattributesforwhichmaybeunknownorunspecified.Movement between places occurred uponways (theterm used by OpenStreetMap) —physical channelssuchas roads, rivers, canals, railways, footpaths,andsealanes—andmayhavebeendirectional. The three types of routes considered here arejourneys,flows,andhistorical_routes: A journey is the record of a specific instance oftravelbyoneormore individuals.Examples include:the 7th century pilgrimage of the Buddhist monkXuanzangacrossChinaand India; the first voyageofCaptainJamesCook,between1768and1771. Aflowistherecordofthemovementofsomething(commodities, people, ideas) between two places,aggregated as a magnitude over a period of time.Examples include: the transport of captive AfricansbetweenWest Africa andBahia in the 17th century;letters between certain correspondents in Paris andPrague in the18thcentury;asourcenetworkof lateNeolithic obsidian artifacts and known sourcelocationsontheAnatolianPlateau. A historical_route asserts a single or compositenamed course of travel between places, takenrepeatedly by unspecified individuals over time,usually forpurposesof commerce.Examples includetheSilkRoadandtheAmberRoutes.Somecorrespondwithnamedroads,forinstancetheViaSalariainItalyisbothawayandahistorical_route.Additional axioms indicated by the relations andcardinalityexpressions(e.g.0…*)inFigure1include:
• Allroutesaresourced,normallytotextualorcartographicdocuments
• The way for a segment (its physical pathdescribedbyageometry)maybeknownandrepresented, unknown, or ignored(Segments with unspecified ways willtypicallybevisualizedasalineorarc)
• Each segment has one or more temporalattribute (“when”), which can be atime_period,(possiblynamed)orasequence(e.g.aftersegmentn)
• Routes and their component segments canhaveanynumberofattributes(properties),dependent upon data sources and projectrequirements
Figure 1. A conceptual model for historical movement
(routes)
The ontology pattern we introduce here isspecialized,ascomparedtohighlevelontologies likeCIDOC-CRM.Wehavenotyetmappedourdistinctiveentities (route, journey, flow, historical_route,segment,when)toexistingontologies.Thetermplaceis commonly found, but usually is synonymouswithlocation; the sense we are adopting is that of thePleiadesgazetteer,but isnot inapublishedontologythatwe'reawareof.Inanycase,wefeelitisbesttofirstlaydownalogicallycoherentsetoftermsandatalaterdateattempttoalignthemwithotherontologies.
Formats TherouteODPhasinformedourdevelopmentandimplementation of recommended standard dataformats. It turnsout all three typesof routes canbeeffectively described in GeoJSON-T, an extendedversion of GeoJSON, the widely-used format forrepresenting geographic FeatureCollections. AFeatureCollectionofrouteswillincludebothPlaceandRoute features.Routesegmentsarearticulatedasanarray of one or more geometries in a route’sGeometryCollection. GeoJSON-T allows optional“when”objects,bothforeachfeatureatthesamelevelas its geometry object and for segment geometries(Figure 2). Features and segments have certainrequiredpropertiesasshown,andcanhaveunlimitedproject-specificproperties.
Figure 2. GeoJSON-T applied to route data
Data To date, seven exemplar datasets have beenconverted from a typical CSV format to GeoJSON-T,usinganewlydevelopedPythonprogram.Threearefor journeys: two by individuals (a 7th centurypilgrimageandamoderncircumnavigation),thethirdby840Venetianshipconvoysinthe13-15thcenturies.Another dataset aggregates those ship journeys asflows havingmagnitudes of journeys and ships. Thelast three are historical_routes: the Roman eraitinerary of the Vicarello Beakers, the route systembetweencourierstationsinMingDynastyChina,andalargesetof“OldWorld”tradeandpilgrimageroutes.
Software The widespread adoption of GeoJSON hasdemonstratedthatforadataformattobeuseful,theremust be software with visualization and analysiscapabilitiesthatsupportsit.Accordingly,anessentialelementoftheLinkedPlacesprojectisdevelopmentofproof of conceptweb software to render GeoJSON-Tdata, forboth routesandplacesalone, toamapandtimelinetogether.Thedevelopmentofthatsoftwareisongoing,andpubliclyavailable.(Figure3).
Figure 3. Linked Places interface (partial view as of March
2017)
Bibliography
Berman,M.L.,Mostern,R.,&Southall,H.(2016).PlacingNames: Enriching and Integrating Gazetteers.Bloomington:IndianaUniversityPress.
Biggs,N.;Lloyd,E.;Wilson,R.(1986),GraphTheory,1736-
1936,Oxford:OxfordUniversityPressChen,P.P.S.(1976).Theentity-relationshipmodel—toward
a unified view of data. ACM Transactions on DatabaseSystems(TODS),1(1),9-36.
Cresswell,T.(2006).Onthemove:Mobilityinthemodern
westernworld.NewYork:Routledge.Gangemi,A.(2005).Ontologydesignpatternsforsemantic
webcontent.InInternationalsemanticwebconference(pp.262-276).SpringerBerlinHeidelberg.
Lowe, J. C., & Moryadas, S. (1975). The geography of
movement.Boston:HoughtonMifflinSimsion, G., &Witt, G. (2004). Datamodeling essentials.
SanFrancisco:MorganKaufmann.
