Wednesday, November 20, 2013

Nomisma, Coins, and SPARQL-generated Maps

Today I'm heading up to the Scholars' Lab at the University of Virginia Library for the fourth (I think?) annual GIS Day festivities.  I'll be discussing the use of Nomisma's RDF and SPARQL endpoint to develop geographic information systems based on linked open data.  Here is my slideshow:


4 comments:

  1. Hello. Great job with Nomisma. I'm wondering, how do you gather the data from different partners? Do you manually ask them to send over their data so that you can convert them to RDF or is there a more automated process? If so, what's the incentive for them to give you data?

    ReplyDelete
    Replies
    1. Any collection that uses Numishare is able export into Nomisma's RDF model out of the box--so that accounts for the University of Virginia collection, Coin Hoards of the Roman Republic, Online Coins of the Roman Empire, and the ANS' own collection. We have a small number of imperial coins from Berlin and all of the British Museum's Roman Republican coins in the Nomisma endpoint. Because the Berlin coins are annotated with RIC ids so consistently, we were able to map to Nomsma URIs, but I had to screen scrape their HTML in PHP to generate RDF since they don't have any export mechanisms. For the British Museum, their Roman Republican Coinage reference numbers contributed to Nomisma, so I was able to use their SPARQL endpoint to generate the RDF dump for Nomisma. Ultimately, we would like to follow examples set by the Pelagios Project, where we accept a VoID RDF file that contains a link to the data dump. Nomisma will parse the dump and perform a consistency check before posting the data into the triplestore. At the moment though, I use command line tools to put the data in, so we don't have a formal, scalable workflow. I'm working on it. We should have something working by the summer.

      We have published the RDF model so that other institutions could join, but our goal is to work with existing collaborators (like the British Museum, Berlin, and Bibliotheque nationale de France) to get some of the major collections into Nomisma. There are a number of incentives for an institution linking their coins to Nomisma-defined concepts. 1. It facilitates multi-lingual interfaces, 2. a sort of controlled vocabulary for terms, 3. coin types are being defined on Nomisma, which reduces the need to manually enter a lot of data, 4. broadens access to their own collection by getting it into the LOD cloud, 5. generally fits the mission of a cultural heritage institution to serve the community--measurements and findspots are used to power visualizations and analyses. Incorporation of Nomisma URIs into archaeological databases would be especially beneficial to publishing excavation data.

      Delete
  2. I love the hyperlinks and maps on the Nomisma pages for mint cities, e.g. http://nomisma.org/id/parium . Is there a way to execute a SPARQL query and have the results formatted like that?

    ReplyDelete
  3. SPARQL results can return XML or JSON, but not necessarily in RDF natively. If you append .'rdf' to a URI (.e.g, http://nomisma.org/id/parium.rdf), you can get the data back. There's also a http://nomisma.org/id/parium.kml if you want the KML file.

    The following query will get all findspots associated directly or indirectly (through a coin type) with Parium.

    PREFIX rdf:
    PREFIX dcterms:
    PREFIX nm:
    PREFIX skos:
    PREFIX geo:
    SELECT DISTINCT ?object ?findspot ?lat ?long ?title ?prefLabel WHERE {
    {?type nm:mint nm:parium .
    ?object nm:type_series_item ?type.
    ?object nm:findspot ?findspot .
    ?findspot geo:lat ?lat .
    ?findspot geo:long ?long
    }
    UNION {
    ?object nm:mint nm:parium .
    ?object nm:findspot ?findspot .
    ?findspot geo:lat ?lat .
    ?findspot geo:long ?long
    }
    OPTIONAL {?object skos:prefLabel ?prefLabel}
    OPTIONAL {?object dcterms:title ?title}
    }

    ReplyDelete

Note: Only a member of this blog may post a comment.