Google Ancient Places

The Google Ancient Places (GAP) project is inspired by two questions: How can I easily find (and view) all the places mentioned in a classical text? And how can I find classical texts about a place I am interested in? GAP is a Google-funded consortium that uses the latest text mining methods to semantically annotate references to ancient places in the Google Books corpus.

How does it work?

We use an adapted version of the Edinburgh Geoparser in order to
deal with the twin problems of toponymic homonyms and synonyms (different places that share names, and single places with multiple
names). This uses a variety of factors, in particular the textual and
spatial clustering of potential locations, in order to identify
 probable place referents. Places themselves are then annotated using the Open Annotation Collaboration (OAC) ontology. The annotations
themselves point to both the online resource – the specific page of a
book, and a unique Web identifier (URI) provided by the Pleiades
Gazetteer for the place itself. Query results will be made
available in RDF and via a simple, RESTful Web service (based on the
Atom Syndication Format), so that results can be related to other
cultural heritage services.

Although the data form GAP’s most significant output we are also
starting to explore ways of visualising it. Dynamic Narrative Maps allow
the user to see the text’s spatial flow when reading it. ‘Placebooks’ list all the known texts that refer to a specific place,
and can also be adapted to show other contextualising information,
such as material culture. Although GAP is predominantly
focussed on Classics, the method is equally applicable to other
 domains.

Indeed, GAP is just one part of the wider Pelagios consortium, led by the Open University and the University of Southampton, which is using the
same web-based framework so that results can be compared across multiple
resource sets, including those from Archaeology, Numismatics,
 Epigraphy and Cartography. In short, GAP is breaking ground that will
allow scholars, students and enthusiasts world-wide to query massive
digital corpora to ask “find me books related to this geographic
location and time period”, and then visualize the results in ways that
help them explore the subject in greater depth.

People

  • Elton Barker, Open University
  • Eric Kansa, UC Berkeley
  • Leif Isaksen
  • Kate Byrne, Edinburgh University
  • Nick Rabinowitz, nickrabinowitz.com