Tuesday, August 18, 2015

Portable Antiquities Scheme in CRRO

The first archaeological materials have been linked into Coinage of the Roman Republic Online (CRRO). Yesterday, Dan Pett began working on linking Republican coins in the Portable Antiquities Scheme to URIs for Michael Crawford's RRC numbers in CRRO. Beginning with just four linked objects yesterday afternoon, in order to iron out the VoID and Nomisma data dump models, the number of PAS coins in CRRO has grown to 66 this morning. In total, we expect more than 1,000 PAS coins in CRRO by the time the database has been fully linked up.

The PAS coins have weights, axes, and diameters that can be used for quantitative analysis. Many have public findspots to show the geographic distribution of types, and of course, most have been photographed. RRC 322/1a, for example, contains links to two PAS coins, one of which has a findspot of Blyth, defined with a URI from the Ordnance Survey gazetteer.

Since most of the PAS Republican coins were minted in Rome, these new data have altered the heatmap distribution for the Nomisma concept of the mint of Rome, with a number of new findspots appearing in England, adding to the two findspots already recorded from Crisis of the Third Century hoards in which a number of University of Virginia Art Museum coins were found.

When the Republican coins in PAS are fully linked into CRRO, we expect that this will enhance the project's value as a research tool for investigating the circulation of Roman coinage into the British Isles.

Incidentally, I updated the underlying SPARQL query for the CRRO Contributors page to display coin counts and publisher information for coins which are not contained in museum collections (since the PAS database contains objects which may have been dispersed to a wide variety of public or private collections). These code updates affect Numishare in general, including OCRE.

Thursday, July 2, 2015

Migrated from OpenLayers to Leaflet in Nomisma, Implemented Heatmaps

I have spent a few days finding solutions to scalability problems loading a lot of points in OpenLayers, and migrated to Leaflet and implemented heatmaps, which reveal context in a more useful way when there are a lot of points. I wrote the following email to the Nomisma list:


We had run into scalability issues with rendering maps of findspots driven by SPARQL for mints in the Nomisma interface. It was not a triplestore issue, but a browser issue. OpenLayers did not handle 1000 points for findspots related to Rome gracefully, so I had to change directions.

First, it should be noted that the the points now reflect unique findspots and not objects (because we can't reasonably map 10,000 coins found in the same hoard, therefore the same findspot).

I switched OpenLayers to Leaflet. I replaced the default KML rendering of points into a heatmap, which is more indicative of distribution overall. See http://nomisma.org/id/rome for example, but you can also see http://nomisma.org/id/sicily for the distribution of (mostly Republican) coins using Sicily as the nmo:hasMint. The IGCH findspots are in the system, so a number of Greek mints have hits as well (http://nomisma.org/id/syracuse).

There are four available base layers. The default is the ancient terrain produced by the Ancient World Mapping Center at the University of North Carolina and hosted by mapbox.com. Also available are modern terrain, OpenStreetMap, and the Roman Empire map.

You can toggle the main point/polygon for the mint/region on or off, as well as the heatmap. You can also turn on the KML layer (which is still available for individual download as well). Leaflet seems to handle a large number of points more efficiently than OpenLayers.

This migration to Leaflet is the beginning of really sophisticated geographic analysis tools that I would like to implement into Nomisma. There are a lot of ways we can make Nomisma data more useful to researchers.

Tuesday, June 23, 2015

Antike Fundmünzen in Europa (AFE) in OCRE

More than 800 coins from Antike Fundmünzen in Europa (AFE), a Römisch-Germanische Kommission (RGK) (a Nomisma partner), have been ingested into the Nomisma triplestore. Of these, only a few dozen are directly linked to URIs in OCRE, but the remaining coins are still available for query by means of directly affiliated triples (such as this SPARQL query below, which gathers a UNION of coins with Vespasian explicitly designated as authority and those that are implied by means of a linked coin type).

SELECT ?object ?type ?weight ?diameter WHERE {
  { ?object nmo:hasAuthority nm:vespasian ; a nmo:NumismaticObject }
  UNION { ?type nmo:hasAuthority nm:vespasian ; a nmo:TypeSeriesItem .
         ?object nmo:hasTypeSeriesItem ?type }
  OPTIONAL {?object nmo:hasWeight ?weight}
  OPTIONAL {?object nmo:hasDiameter ?diameter}

These are the first archaeological materials ingested in Nomisma, each with findspots.

See http://numismatics.org/ocre/id/ric.2_1(2).ves.520, for example, of a Vespasianic coin type with a findspot and measurements from the AFE database.

Thursday, June 18, 2015

Berlin coins updated in OCRE and CRRO

The Berlin LIDO/XML export data for Roman Imperial Coinage references has been reprocessed, and coins from the Münzkabinett from Valerian through Quintillus are now available in OCRE. The number of imperial coins from Berlin grew about 150 to 3,749. Additionally, the RRC export was reprocessed into RDF following the new Nomisma ontology, and the total number of Republican coins from Berlin available in CRRO has grown to 1,080. The ANS coins for RIC V were updated yesterday, and the University of Virginia coins just before that. I expect to update the RIC V coins from the British Museum in the near future. This will add over 2,000 more coins to OCRE.

Presently, nearly 35,000 coins from the ANS, British Museum, Berlin, and the University of Virginia are available in OCRE. I expect the first batch of archaeological materials from Antike Fundmünzen in Europa (AFE) to be ingested into the Nomisma triplestore soon. This is a tremendous milestone.

Wednesday, June 17, 2015

All RIC types through Quintillus published in OCRE

Last week, we published all of the types from Valerian and Gallienus in OCRE. A few days ago, we added Claudius Gothicus and Quintillus. There is a subtle difference between the types from the reigns of Valerian and Gallienus and the emperors after 268 in our RIC V workflow. Due to the tremendous number of subtypes from Valerian and Gallienus (and subtypes of these subtypes, when considering every permutation of obverse portrait and mint mark), we decided we needed to speed up the publication process. Beginning with Claudius Gothicus, only the parent type numbers will be published to OCRE so that we will complete RIC Volume X by the close of the NEH grant two years from now. If/when we have the time, we will return to RIC V to fill in the missing subtypes.

Capturing the parent types is of the utmost importance, since most museum and archaeological databases will likely have designated their coins as a parent type, e.g. Gallienus 210, and not Gallienus 210k or Gallienus 210k Subtype 1. The ANS has gone into this level of granularity with respect to linking coins to OCRE subtypes (as have I with linking the University of Virginia collection), but it is doubtful anyone else has. Indeed, at first glance, neither Berlin nor the British Museum link to the subtypes, many of which were not distinctly numbered within RIC. [As an aside, Mantis and UVa have been updated to link to OCRE, but BM and Berlin coins are not yet available.]

However, in order to support the traversal from parent types to subtypes, I have made some enhancements to the OCRE UI (which are not inherent to Numishare). It is possible to link directly to Gallienus 210 and have those coins show up under "Examples of this Type" in the parent record page, but coins linked more granularly to subtypes will also show up under the "Subtypes" section underneath. Click on the Gallienus 210, 210k, and 210k.1 above for an example. What this means is that an archaeological database can link to the parent type when the legend or mint marks are not legible enough for precise identification, but the numismatic data and images can still be made available through OCRE.

In semantic web terms, subtypes are linked to parent types via the skos:broader property. SPARQL queries have been updated throughout Numishare to infinitely gather subtypes with the skos:broader+ modifier, which is enabled by default in the current version of Fuseki we employ as the Nomisma.org SPARQL endpoint (http://bit.ly/1N2pe81).

Wednesday, May 13, 2015

Monday, May 11, 2015

Clarifying licenses in Mantis

As many of you might know, several ANS projects, like OCRE and CRRO, already have well-defined Open Database Licenses. In the ANS database itself, Mantis, we had a generic copyright and terms of use statement. We have moved forward with clarifying the licenses on Mantis. The data are made freely and openly available with an Open Database License. The images themselves are now under the Creative Commons Attribution/Non-Commerical license. People are free to link to our coin images from external sites and projects. While this may have been going on anyway, we have now formalized the right to do this.