Numismatic News: Network graph visualizations for Hellenistic monograms

Network graph visualizations for Hellenistic monograms

I have recently unveiled a new data visualization feature in Numishare that has the broadest impact in the Hellenistic Royal Coinages project, with its thousands of monograms: a network graph of the relations between a monogram and associated monograms that appear on the same types. This is formed by a SPARQL query (below) that is serialized by XSLT into the JSON model for the d3plus Network vis. A secondary iteration of queries is executed to generate an additional level of nodes related to each monogram associated directly with the root URI. In theory, you could iterate beyond this secondary relationship, but it's probably unnecessary for the purposes of a simple visualization. These data could be loaded into a desktop network analysis tool such as Gephi for more sophisticated display.

SELECT * WHERE {
  BIND(<http://numismatics.org/pella/symbol/monogram.price.829> as ?symbol)
  ?type nmo:hasObverse/nmo:hasControlmark|nmo:hasReverse/nmo:hasControlmark ?symbol .
  ?symbol skos:prefLabel ?symbolLabel ;
          crm:P165i_is_incorporated_in ?image .
  ?type nmo:hasObverse/nmo:hasControlmark|nmo:hasReverse/nmo:hasControlmark ?altSymbol . 
          FILTER (?altSymbol != ?symbol).
  ?altSymbol skos:prefLabel ?altSymbolLabel ;
          crm:P165i_is_incorporated_in ?altImage .        
}

Visualization of Price monogram 222

This network graph is a novel approach to investigating the potential meanings of these symbols, as it allows for the exploration of patterns that were never previously observed, at least at the scale that we are able to present in the HRC project. The thickness of the edges is dependent upon the number of types that share relationships. Therefore, monogram 769 and 55 share numerous types. I am still working on making some stylistic tweaks to the display.

There are two major areas of completeness that will be addressed in the future:

1. In the process of developing this visualization, it has become apparent that there are numerous gaps in the typology where monogram URIs have not been inserted into the Price spreadsheet (and perhaps the Ptolemaic and Seleucid ones as well), so the monogram network visualization isn't necessarily a full accounting of all relationships.
2. We have not yet gone through the monograms of the three major corpora to link the same symbol together into a cohesive, distinct set of monogram URIs. We know that the same monograms appear in Price on Alexanders as those that appear on the coinages of the Seleucids and Ptolemies, but they are not joined together. Therefore, the network visualizations in PELLA don't necessarily include monograms from Seleucid Coins Online or Ptolemaic Coins Online, unless they overlap with a small selection of early Seleucid or Ptolemaic types that were struck in the name of Alexander the Great.

One other recent enhancement is the implementation of a GeoJSON serialization for a monogram (appending .geojson to the monogram URI or requesting the content-type 'application/vnd.geo+json' via HTTP content negotiation). This GeoJSON is formed by three distinct SPARQL queries to get the mints, hoards, and findspots associated with a monogram URI. While this map did exist already, the enhancement includes the numeric counts of mints or findspots, which are stylized into differently-sized bubbles in Leaflet.

Price Monogram 1090 is one of the geographically best-represented symbols, particularly prominent in East Greece and the Black Sea. Adding a histogram of issue dates would be a useful tool in the future.

The next phase of development will be to include distribution visualizations in order to generate charts that show the numbers of denominations, mints, authorities, etc. that issued types that depict these monograms.

Posted by Ethan Gruber at 7:38 AM