Encoding and querying historic map content

  • View
    796

  • Download
    3

  • Category

    Science

Preview:

DESCRIPTION

Slides courtesy of Simon Scheider. Presentation for our paper at AGILE 2014: Simon Scheider, Jim Jones, Alber Sanchez and Carsten Keßler (2014) Encoding and querying historic map content. In Joaquín Huerta, Sven Schade, Carlos Granell: Connecting a Digital Europe Through Location and Place. Springer Lecture Notes in Geoinformation and Cartography 2014: 251–273. DOI:10.1007/978-3-319-03611-3_15

Citation preview

http://lodum.de

Encoding and querying historic map content

Simon Scheider*, Jim Jones*, Alber Sanchez*, Carsten Keßler§

*University of Münster, Institute for Geoinformatics, Münster§Hunter College, Department of Geography, NY

How can (we support) historians (in) find(ing) (answers in) maps?Question:“What was the type oflandcover around Hildesheimin the 19th century?”

1) Manual search(through 20.000 maps?)

2) Text field search:- title: (“Gaußsche Landesaufnahme”“Berghe Ducatus”,...)- author:(Gerhard Mercator, ...)- year of production(1680, 1839, ...)-key words: (“topographic map”, “Flurkarte”)

Sample from the map repository at ISTG (Institute for comparative urban history), Münster

How can (we support) historians (in) find(ing) (answers in) maps?

Technical challenges:

1) Manual search(through 20.000 maps?)

2) Text field search:- title: (“Gaußsche Landesaufnahme”“Berghe Ducatius”,...)- author:(Gerhard Mercator, ...)- year of production(1680, 1839, ...)-key words: (“topographic map”,“Flurkarte”)

Not scalable!

Language?

How to pick the „right“ terms? (which correspond

to the answer?)

How to pick the „right“ place/space? („the area

around Hildesheim“)

How to pick the right time? („19th century“)

There are many languages in maps (Latin, ...)!

Placenames are changing! Historic maps are distorted and lack CRS!

Terms are ambiguous!There is too much content!There is nameless content (e.g. „landcover around Hildesheim“)!

How can (we support) historians (in) find(ing) (answers in) maps?More questions:“What was the extent of Prussia?”“Which territories were part of Prussia?”“Which Prussian territories were acquired by Friedrich-Wilhelm of Brandenburg, thegreat elector?“

Answer depends on time ... and ambiguity of names ...

Prussia 1806

„Brandenburg“ (Prussia) 1688

How can (we support) historians (in) find(ing) (answers in) maps?A map answering detailed historical knowledge:“How many people did Napoleon’s army have when soldiers arrived in Smolensk during

his 1812 campaign?““What were thelowest temperatures during Napoleon’s campaign?”“Which places did Napoleon’s army come across during the 1812 campaign?”

Minard’s map about Napoleon’s invasion of Russia 1812:

How can (we support) historians (in) find(ing) (answers in) maps?

Research topics we addressed in the paper:

1) How to precisely encode and query - semantic, - spatial and - temporal map contents?

2) How to deal with - wealth of content- language/naming ambiguity?

Linked spatio-temporal data for historic mapsLinked spatio-temporal data enables

1. a simple und universal approach to describe semantic contents of (map) documents (namely, a graph)

2. complex content queries (beyond text search) using diverse languages

3. logical expressions and reasoning for approximate content descriptions/queries

4. linking to external resources (URI) ...and therefore: (re)-using resources and crowdsourcing

5. using spatial (OGC simple feature) and temporal references

Map

Berghe Ducatus

Gerard Mercator

is acreator

coordx: ….y: ….

mapsArea

“1550”mapsTime

Berg

State

is aKöln

maps

City

“1512”

birthDate

is a

Formally encoding map contentsMap contents can be treated as sets of assertions that can be extracted by looking at the map:

In the Semantic Web, - nameless content- wealth of content can be addressed by intensionality:- logical quantification- blank nodes

In linked data, this translates into a named graph:

Vocabularies we reused:

-For map area as well as content space:GeoSPARQL ontology (prefix geo):

OWLtime (prefix time):

- For document properties:...

Vocabularies for historic map contentsMaps as documents (prefix maps) :http://geographicknowledge.de/vocab/maps

Vocabularies for historic map contentsContent phenomena (prefix phen):http://geographicknowledge.de/vocab/historicmapsphen [.rdf/.jpg]:

(reuse ofany geographic/historicalontology, such as: )

Encoding maps as linked dataFor example, the map about Hildesheim 1840:

Document (graph) represents Content graph (describing the map as document) (describing content assertions)

Georeferencing and annotating historic mapshttp://data.uni-muenster.de/georeferencer/georef.html1) Georeferencing map image:using control points(known locationsin Open Street Map)

Georeferencing and annotating historic maps

2) Determinemap window

Automaticcalculationof- map scale- map area

Georeferencing and annotating historic maps

3) Describedocument:- time- creator- size- document URL…

Georeferencing and annotating historic maps

4) Describe contentsAutomaticallysuggested content based on map area,time window

Reuse of externalinformation recources(e.g. the state Bergat Dbpedia)Different historianscan contribute tothe same map

Publishing maps and their contents

RESTful publication (accessible over http): - As RDF files or KML files- over SPARQL endpoint- Can be accessed over the Webfor display or search

Display with Google Earth:

Querying historic map contentsWhich maps contain information about ...

Querying historic map contentsWhich maps contain information about ...

Querying historic map contentsWhich maps contain information about ...

Conclusion• Historic map contents =: named RDF graphs!• ... allows the expression of map contents in a precise way as a set of (triple) assertions•... and the linking of maps as documents with their contents, reusing published vocabularies and external links (Dbpedia)• ... enables crowdsourcing of content descriptions• ... makes possible intensional descriptions (using blank nodes) in order to cope with the wealth of content, nameless content and content approximation • ... enables retrieval of maps that can answer historians‘ questions!

Future work

• Tools that help encoding map contents for non-trained users (beyond georeferencer)?• Tools that allow non-trained users to formulate (visual) content based queries?

>SPEX (spatio-temporal content explorer)(see https://github.com/lodum)

Dr. Simon ScheiderInstitut für Geoinformatik derWestfälischen Wilhelms-UniversitätHeisenbergstraße 2, 48149 Münstersimon.scheider@uni-muenster.deTel.: 0251 I 83-30088

Thanks for your attention!