CHGIS: from Historical GIS to Open API

Merrick Lex Berman

Center on Religion and Chinese Society

Purdue University, 27th Sep 2016
General goals of Digital Humanities

access to cultural / historical information

tools to find, browse, mashup, query, map, and analyze this info

communicate the resources, topics, and findings of digital scholarship

utilize digital resources and methods for teaching and learning

connect communities (across disciplines, across sectors, and the public )

 Trend of Digital Humanities projects

access to cultural / historical information

tools to find, browse, mashup, query, map, and analyze this info

communicate the resources, topics, and findings of digital scholarship

utilize digital resources and methods for teaching and learning

connect communities (across disciplines, across sectors, and the public )

movement from

introverted projects by individual scholars ->

to collaborative cross-disciplinary projects ->

to exposing data and methods to wider academic communities & the public.
Challenges of Digital Humanities

how to bridge the gaps between

(a) primary sources

(b) scholars with domain expertise
(c) technical implementations

http://www.clir.org/pubs/reports/pub151/pub151.pdf
 Development and Publication of CHGIS

Use cases for geocoding Buddhist sites

Temporal Gazetteer and Linked Open Data

Source texts for Chinese Local Gazetteers
Compiled into Dynastic Gazetteers
Historical Atlas of China ( volumes)
Requirements for the CHGIS Data Model (2001)

find admin units in existence at a particular time

search by placename

filter by admin status (feature type)

determine the administrative hierarchy for a particular place instance

show how a particular admin unit changes over time

 How CHGIS was compiled

Maximize Point Locations

Record each change of Placename, Unit Type, or Location

Digitize changing boundaries based on printed maps annotated by scholars

For each set of changes, detailed source citations are stored

Relational Database for administrative hierarchy, sequence in time, sources

 How the 1911 county boundaries were drawn

Working from a medium resolution photograph (not a scan) of each county map

Maximum number of town points are digitized based on known current locations

The town locations, natural features, & modern county boundaries are used to determine the 1911 boundary

Historical map sources for 1911 Modern basemaps used for digitzation
counties from late Qing atlas sheets of points and polygons
After compiling all the counties: 1911 Time Slice
 Compiling the Time Series boundaries

Time Series of boundary changes

Annotations of boundary changes are keyed to color shading and time period notes

Printouts of reconstructed 1911 boundaries are marked up by editors

 Compiling the Time Series points

Locations of the changing county, prefecture, province capital seats are recorded

Printouts of modern basemaps are marked with dated points by editors

CHGIS database of all historical instances
CHGIS relational database (circa 2003)
CHGIS Version 1 (2001)
CHGIS Version 2 (2003)
CHGIS Version 3 (2005)
CHGIS Version 4 (2007)

Xian 1980 Projection

CHGIS Version 5 (2012)

~40,000 towns & villages

 Time Series data - CHGIS Version 5 (2012)

~3,000 prefecture polygons

~3,500 prefecture capitals

~10,000 county seats

Time Series data - CHGIS Version 6 draft (2016)

v5

v6
CHGIS Datasets Publication and Archiving

CHGIS website(s) DataVerse archive

Shapefiles Shapefiles

ACCESS database ACCESS database

DEM and Supplements DEM and Supplements

Scanned Historical Maps

Reports and Documentation

http://www.fas.harvard.edu/~chgis/

http://yugong.fudan.edu.cn/ https://dataverse.harvard.edu/dataverse/chgis
User Base Managing Expectations

Steep learning curve

False assumptions based on other common geo-spatial technologies

False Expectation: GIS = Google Maps
Webmaps are very different from Vector GIS

Cannot download and re-use Downloadable but very time-consuming

Web map platform can bring users to GIS
Vector GIS and WorldMap ecosystem
Delivering raw data vs. Defined layers

ChinaX
Shapefiles for download Chinas History in Maps
Use cases for geocoding Buddhist sites
Source for Qing Geography

(Da Qing yitongzhi)

[Qing Dynasty National Gazetteer]
Jiaqing (560 juan) Beijing, 1842

11 v. (7036 p.), Taipei, 1934

 Example from

page from Yitongzhi

Index version of Buddhist Temples

Chugoku no jiin.
[Temples of China]
Yajima, Genryo, editor.
Thoku Daigaku Fuzoku Toshokan , Sendai, 1966. [1941] 264 p.
Comparison of Buddhist Temples text
Geocoding Temple locations with CHGIS data
Geocoding Temple locations with CHGIS data
Geocoding Temple locations with CHGIS data
Geocoding options - planimetric accuracy

location for new data is

snapped to a known point

known point
Geocoding options - planimetric accuracy

location for new data is

location for new data is calculated based on
snapped to a known point distance and direction
from a known point

known point known point

Geocoding to calculated locations: benefits
Geocoding to calculated locations: drawbacks
Mapped locations of all 2,400 temple sites
COUNT density of temples at geocode locations
 Source for Qinghai Temple Locations

(Gan Qing Zang Chuanfojiao Siyuan) [Buddhist

Traditional Monasteries in Gansu, Qinghai, and Tibet] Xining, 1990.
Data Entry of Qinghai Temple Locations
Distribution and COUNT of Qinghai Temples
Comparison of Yizhongzhi and Qinghai Temples
Greater Tibetan region monastery sites
Heat map of Tibetan sites and Yitongzhi sites
Heat map of Qing exams & Ming courier routes
Ming Courier Routes Network Model

http://maps.cga.harvard.edu/chinapostoffice/
Animation of movement -- Biographical Network

http://www.fas.harvard.edu/~chgis/work/docs/papers/ouyi_flash_demo.swf
 Animation of events -- Ming Examinees

https://www.youtube.com/watch?v=LLLchSgVcXo
Temporal Gazetteer and Linked Open Data
 GIS vs. Gazetteer: breaking the bottleneck

Using GIS boundaries Using gazetteer entries

- extremely time consuming - enter data based on text descriptions
- limits to what boundaries are known - relations are not based on geometry
- interpretations very hard to document - attestations are easy to document
- missing parent boundaries leave gaps - boundaries are not a bottleneck
CHGIS Gazetteer XML Webservice (2006)
CHGIS XML to Temporal Gazetteer (2013)

Existing System Proposed System

CHGIS Gazetteer Web Service Temporal Gazetteer Web Service

Temporal Gazetteer System Architecture (2014)
TGAZ API launched (2014)
TGAZ test: Russian Gazetteer (2014)
TGAZ RDF interchange format (2014)
TGAZ RDF interchange format (2015)
 TGAZ RDF ingested into Pelagios (2015)

Dump of database to RDF Ingested gazetteer

71,000 records Into Pelagios system

http://pelagios.org/recogito/

<http://chgis.hmdc.harvard.edu/placename/hvd_1> a lawd:Place ;
rdfs:label "Ba Zhou"@en ;
lawd:hasName [ lawd:primaryForm ""@zh-Hant ] ;
lawd:hasName [ lawd:primaryForm ""@zh ] ;
lawd:hasName [ lawd:primaryForm "Ba Zhou"@en ] ;
geo:location [ geo:lat 39.10154 ; geo:long 116.39525 ] ;
gn:countryCode "cn" ;
dcterms:description "" ;
dcterms:temporal "start=1820; end=1820;" ;
dcterms:subject "department " ;
.
<http://chgis.hmdc.harvard.edu/placename/hvd_2> a lawd:Place ;
rdfs:label "Zhenghuangdengsiqi Muchang"@en ;
lawd:hasName [ lawd:primaryForm ""@zh-Hant ] ;
lawd:hasName [ lawd:primaryForm ""@zh ] ;
lawd:hasName [ lawd:primaryForm "Zhenghuangdengsiqi
Muchang"@en ] ;
geo:location [ geo:lat 41.21478 ; geo:long 113.89422 ] ;
gn:countryCode "cn" ;
dcterms:description "" ;
dcterms:temporal "start=1820; end=1820;" ;
dcterms:subject "pasture land " ;
.
Text Annotation in Pelagios / Recogito (2015)
Map Annotation in Pelagios / Recogito (2015)
Integration of CHGIS data with MARKUS (2015)
Integration of CHGIS data with MARKUS (2015)
TGAZ JSON format (2014)
 TGAZ JSON for future Editing Interface (2015)

JSON gazetteer record

{ "system" : "China Historical GIS, Harvard University and Fudan

University", "license" : "c. 2014", "uri" :
"http://chgis.hmdc.harvard.edu/placename/hvd_80547", "sys_id" :
"hvd_80547", "sys_id of alternate" : "", "spellings" : [ { "traditional
Chinese" : "", "exonym language" : "", "attested by" : "", "note" :
"" }, { "simplified Chinese" : "", "exonym language" : "", "attested
by" : "", "note" : "" }, { "transcribed in Pinyin" : "Tengchong Fu", "attested
by" : "", "note" : "" } ], "feature_type" : { "name" : "", "alternate name" :
"", "transcription" : "fu", "English" : "superior prefecture" }, "temporal" :
{ "begin year" : "937", "begin rule" : "2", "end year" : "1273", "end rule" :
"4" }, "spatial" : { "object_type" : "POINT", "xy_type" : "point", "latitude" :
"25.05579", "longitude" : "98.50229", "source" : "FROM_FD",
"present_location" : [ { "country code" : "cn", "text" : "
", "source" : "Fudan", "attestation" : "" } ] }, "historical_context" :
{ "part of" : [ ], "preceded by" : [ ] }, "data source" : "CHGIS", "source
note" : "9371273
}

Revisions will
Be written to
TGAZ
database
 Points for consideration

Geodatabase is only useful to GIS users

Publishing in GIS formats limits the potential user base

Publishing in web-based applications has wider audience

Web-based applications are siloes

Open APIs provide machine-actionable access to content

APIs and LOD will outlive any web-based applications or portals

TGAZ Project Resources

TGAZ API: http://chgis.hmdc.harvard.edu/tgaz/

Papers: http://www.fas.harvard.edu/~chgis/gazetteer/

Merrick Lex Berman: mberman@fas.harvard.edu