Tuesday, March 24, 2020

First batch of Philip II coins added to PELLA

The first few collections of Philip II coins have been linked to new Le Rider URIs and have been made accessible in PELLA through Nomisma.org's SPARQL endpoint. The American Numismatic Society linked 656 coins to these new URIs. Over the weekend, Karsten Dahmen at the Berlin Münzkabinett cataloged the first 17 coins of their collection. Additionally, I used Matt Lincoln's old data dump of years-old British Museum RDF data running locally in an endpoint to query for Le Rider references among their coins. Using OpenRefine to perform some additional cleanup, almost 60 coins of Philip II from the British Museum are now available in PELLA. Furthermore, just about a half hour ago, Julien Olivier at the Bibliothèque nationale de France sent a spreadsheet of 202 coins of Philip II with their corresponding Le Rider URIs. At this phase, there are now over 900 physical specimens linked into these typologies in PELLA. Not bad since we only just published them less than a week ago!

PELLA Philip II 137 is one of the best-represented types.

I expect a lot more contributors to come along as the cataloging process begins to proliferate through partners such as KENOM and NUMiD.

As it stands now, there are 382 total parent types in PELLA. The coins from the aforementioned collections are cataloged with a mixture of Le Rider (subtype [skos:broader] or skos:exactMatch [for plate-figure numbers]) URIs and PELLA parent type URIs. With the following SPARQL query, I can get a count of the number of photographed specimens per type:

PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
PREFIX dcterms: <http://purl.org/dc/terms/>
PREFIX foaf: <http://xmlns.com/foaf/0.1/>
PREFIX nm: <http://nomisma.org/id/>
PREFIX nmo: <http://nomisma.org/ontology#>
PREFIX skos: <http://www.w3.org/2004/02/skos/core#>

SELECT ?type (count(?coin) as ?count) WHERE {  
  ?type dcterms:source|skos:exactMatch/dcterms:source nm:pella_type_series .
  ?coin nmo:hasTypeSeriesItem|nmo:hasTypeSeriesItem/skos:broader ?type .
  {?coin nmo:hasObverse/foaf:depiction ?img}
  UNION {?coin foaf:depiction ?img}
} GROUP BY ?type

Hoards of Philip II coins
There are 258 types of Philip II with at least one photographed specimen, which is about two-thirds of the entire range of coinage. Not too bad for just several collections working together to catalog their coins in a few days.

Now that we have linked the Philip II Nomisma ID to typologies and specimens, some of these specimens also include hoard references, therefore populating the Philip II URI with additional research context for geographic and statistical visualization.

Average weights of tetradrachms from Philip to Alexander

As you can see from the chart above, the weight of Macedonian tetradrachms remains consistent from Philip II to Alexander until 336 B.C (https://bit.ly/2wCuH8G).

Thursday, March 19, 2020

The gold and silver coinage of Philip II published to PELLA

After considerable effort by ANS curator Peter van Alfen, the gold and silver coin types of Philip II of Macedon have been published to PELLA. This typology, based on Georges Le Rider's 1977 Le monnayage d’argent et d’or de Philippe II frappé en Macédoine de 359 à 294, has been numbered 1-382. Le Rider's corpus is actually a die study, and the numbering system is based on die combinations rather than types. As such, there are about 2,200 or so die combinations that correspond to the 382 types. These types were given a new numbering scheme, pella.philip_ii.1 to 382, but all of the Le Rider numbers are also URIs in order to establish a concordance between Le Rider and the new scheme so that collections that cataloged their coins with Le Rider numbers can submit their RDF with those URIs or map the Le Rider to the PELLA type number.

Le Rider numbers as subtypes or deprecated types

The correspondence between PELLA and Le Rider numbers is either 1:n or 1:1. When a typology has multiple possible die combinations, the Le Rider number is considered to be a subtype of the PELLA super type. For example, there are three die combinations for PELLA Philip II no. 1, Le Rider 1.1 to 1.3. If a museum has cataloged one of their coins to Le Rider 1.1, that coin will show up on the subtype page for that combination as well as the page for PELLA Philip II 1, which gathers all of the physical specimens linked directly to that URI or any possible subtype URI (via Nomisma.org's SPARQL endpoint).

Philip II 125 with the first specimen from a Nomisma partner.

The other category in which a PELLA and Le Rider number might correlate is 1:1. In these cases, the Le Rider URI still exists, but is not a subtype (linked in the RDF as skos:broader). Instead, the Le Rider Linked Open Data is dcterms:isReplacedBy the PELLA URI, which forces an automatic semantic HTTP 303 redirect in the browser (e.g., Le Rider 5.120). The underlying RDF for the Le Rider URI is still accessible through content negotiation or appending .ttl, .jsonld, or .rdf onto the URI. These two URIs are still linked together by skos:exactMatch, which facilitates the display of coins linked to Le Rider URIs on the PELLA page.

Uploading spreadsheets in Numishare

The publication of this typology represents a breakthrough in another way. Last summer, I spent several weeks developing a spreadsheet import mechanism in the Numishare back-end. I detailed it here (along with recommendations for structuring numismatic data in Google Sheets). After some further tweaking this week, the Philip II typology is the first spreadsheet data imported into Numishare in production. Out of all of the typology projects we have published online in the last 8 years (since the OCRE prototype developed in 2012), this is the first one that did not require me to write an intermediate PHP script. I cannot overemphasize how important this is. Curators can now formulate their own data, according to the specifications above, and publish new projects without technical intervention. After the conclusion of the Hellenistic Royal Coinages project in May, I will turn my attention to refactoring older spreadsheets from OCRE into this new format, and major updates in OCRE can be made directly by curators. I have essentially coded myself out of a repetitive task, saving myself a lot of time and the ANS a lot of money in the long term.

What's next for PELLA?

Now that the URIs for Philip II are activated, I imagine that our colleagues in Berlin and Paris will begin cataloging with them. Peter has updated our curatorial database with these new IDs, but the data have not been pushed from our [terrible] FileMaker database to Mantis, but you should expect to see the ANS's coins of Philip II online in the coming days. In the mean time, a single tetradrachm from the Fralin Museum at the University of Virginia is the first contributor to the new corpus, for Philip II 125.

Tuesday, March 10, 2020

About 800 "Greek" concepts published to Nomisma

After two weeks of solid research and data entry labor by Andy Meadows and I (and after review by the Greek committee), about 800 new or updated concepts corresponding to the Greek world have been published to Nomisma.org in three spreadsheets. These entities include people (rulers, usually) and their associated dynasties and corporate entities. Aside from people and organizations we typically consider "Greek", these lists include entities from other domains that have always been historically part of the study of broader Greek coinage, including Parthian, Indo-Scythian, Numidian, Arabian, etc. These entities aren't a comprehensive listing of every possible ruler that issued coinage from a particular realm, but form a large portion of these rulers, reflecting the combination of an older spreadsheet of rulers with the entities necessary for the publication of a new Inventory of Greek Coin Hoards database (as part of the NEH-funded Hellenistic Royal Coinages project). Corporate entities can be separated into their own spreadsheets for further revision by subject specialists in order to fill in gaps.

What's useful is that while we are able to use the W3C Org ontology to link people to their roles in larger corporate entities (and the start and end dates of their reign(s)), we are also able to apply the same org:Membership model to link one foaf:Organization to a larger one. That is to say, the Kingdom of Cimmerian Bosporus was independent from 438-107 B.C., but from 107 to 63 was part of the Kingdom of Pontus, and from 63 B.C. to 370, operated as a client-kingdom of Rome.

This enables us to execute queries for all of the lesser kingdoms that served as client-states of Rome:

  ?org a foaf:Organization ;
         org:hasMembership/org:organization nm:roman_empire ;
         skos:prefLabel ?label FILTER (langMatches(lang(?label), "en"))

This can be taken a step further to list the client-kings of Rome by means of the link between the ruler and their kingdom and dates of their reign and filtering it against the date range in which that kingdom was a client-kingdom of Rome.

SELECT ?person ?p_label ?p_start ?p_end ?org ?label WHERE {
?org a foaf:Organization ;
         org:hasMembership ?membership ;
         skos:prefLabel ?label FILTER (langMatches(lang(?label), "en")).
  ?membership org:organization nm:roman_empire ;
              nmo:hasStartDate ?client_start;
              nmo:hasEndDate ?client_end .
  ?person org:hasMembership ?pMembership ;
          skos:prefLabel ?p_label FILTER (langMatches(lang(?p_label), "en")).
  ?pMembership org:organization ?org ;
               nmo:hasStartDate ?p_start ;
               nmo:hasEndDate ?p_end .
  FILTER (?p_end > ?client_start && ?p_end < ?client_end)
} ORDER BY ?org ?p_start

The query above results in the following table of rulers sorted chronologically. The first client-king of Bosporus is Pharnaces II, the son of Mithradates VI, who took over in 63 B.C.

Now that the publication of these entities is complete, I will turn my attention back to reconciling authorities in IGCH data in OpenRefine. The new coin hoard database should be ready by the end of March, and it will facilitate new modes of query that include querying by dynasty and corporate entity by means of the person-org/dynasty relationship inherent in Nomisma's LOD data model. Greek numismatists will finally have a decent open access tool for coin hoard research, coupled with the publication and interlinking of our archival records in Archer.