Coin Hoards of the Roman Empire joins the Linked Open Data cloud

In a significant enhancement to the study of Roman imperial coinage, nearly 5,000 coin hoards from the Oxford Coin Hoards of the Roman Empire project have been integrated into the numismatic Linked Open Data and are accessible in Online Coins of the Roman Empire. Nearly 5,000 coin hoards have been uploaded into Nomisma's SPARQL endpoint linking to about 14,000 Roman imperial coin types. This represents nearly one-third of all hoards in the database; the only hoards integrated into the LOD cloud are those with at least one OCRE URI, although it would be possible to upload all hoards in the future, with references to Nomisma URIs for authorities, denominations, etc.

Many CHRE hoards include geographic coordinates linking to the known findspot place name, although these places had not been coreferenced to URIs in a gazetteer system, such as Geonames. Prior to tranforming the CHRE export into RDF, I used OpenRefine to interact with Geonames APIs to link to the lowest-level geographic entity referenced by the CHRE record, whether it is a city, region, or country. A handful of places do not have Geonames URIs, but link to Wikidata.org instead. Upon uploading the RDF data into Nomisma.org, the Geonames URIs were linked to Wikidata, and the geographic hierarchy for findspots was also extracted from Wikidata and uploaded to Nomisma. It is therefore possible to query all hoards found within Greece by querying based on the Wikidata or Geonames URI for Greece, regardless of the source dataset. This enables more complex queries across datasets, such as denarius hoards of the late Roman Republic and early Principate, combining the hoards of Coin Hoards of the Roman Republic and Coin Hoards of the Roman Empire. 

In addition to significantly enhancing the geographic distribution maps depicted on the coin type pages in OCRE, the distribution maps of related Roman imperial numismatic concepts on Nomisma will also show a large number of hoards. It should also be noted that single finds of gold coins are recorded in the database, and so the geographic distribution of aurei and solidi are greatly improved.

The distribution of the coinage of Hadrian, from Hadrian's Wall to India

The distribution of Augustus 206, an aureus

Once ingested, the Nomisma-backend will extract the geographic hierarchy from Wikidata, making it possible to query for all hoards found within Romania, for example:

 

SELECT * WHERE {
  ?hoard a nmo:Hoard ;
           nmo:hasFindspot/crm:P7_took_place_at/crm:P89_falls_within ?findspot .
  ?findspot crm:P89_falls_within+ <http://www.wikidata.org/entity/Q218> ;
            rdfs:label ?label
}

Many thanks to Marguerite Spoerri Butcher for providing the export from CHRE.

Excavation coins added to Levantine Coins Online

About 60 coins with findspots in Israel (50 of which are from excavations) have been added into Levantine Coins Online. The excavation coins are from KAMAT (Staff Officer for Archaeology – Civil Administration of Judea and Samaria) and the Israel Antiquities Authority, and the remaining coins are from private collections.

Yehud type 24 has the largest representation of physical specimens, with a handful from excavations.

Yehud 24 with findspots

The findspots have also propagated into the broader numismatic network graph in Nomisma.org, with findspots visible in the maps of individual concepts relating to Yehud coinage.

Distribution of finds of Yehud Medinata

ANS Latin American department updated in Mantis

Thanks to the hard work of Sami Norling, our contract data cleanup specialist, the Latin American department at the American Numismatic Society has been migrated from our 20 year old FileMaker Pro database to CollectiveAccess. The department consists of 32,000 objects, more than 29,000 of which are publicly accessible on Mantis. The data normalization process has significantly improved the consistency and quality of the department, with the majority of place names linked to Geonames.org URIs, making it possible to plot production places and a limited number of findspots on the Leaflet-based map interfaces. A significant number of people, corporate bodies, and dynasties have been linked to Wikidata.org, enabling the standardizing of names internally within the department and with similar objects in other departments (for example, there is substantial overlap in artists, issuers, and authorities from Medals and Decorations).

 

A map showing the distribution of production places and findspots of the Latin American department.

Students, scholars, and collectors of the coinage of Central, South America, and the Caribbean will find these improvements extremely useful. The integration of Geonames and, particularly, Wikidata will enable us to build the next generation of query and visualization features in Mantis once we complete the database migration process. The last remaining departments are East Asian, Medieval, and Modern European.

Harvard Art Museums added to BIGR

Thanks to the work of Simon Glenn (Ashmolean Musem, Oxford) and Laure Marest (Harvard Art Museums), 19 coins from Harvard Art Museums have been added to the Bactrian and Indo-Greek Rulers (BIGR) project. HAM is the 10th contributor to the project, which now contains links to more than 5,700 specimens.

A Harvard coin depicted at Euthydemus I 2.15

A spreadsheet of Harvard -> BIGR URI concordances and some basic metadata allowed me to load the spreadsheet into OpenRefine in order to extract the IIIF image and manifest URIs and export the resulting data directly into Nomisma-compliant RDF using a template. This is a departure from earlier iterations of the workflow which necessitated writing a PHP script to harvest data from the API and attempt to automate the matching of type references to URIs.

New Nomisma project partners are increasingly creating their own RDF/XML exports for ingestion into the Nomisma SPARQL endpoint or providing spreadsheets already embedded with coin type URIs, enabling a more seamless transformation into RDF through minor OpenRefine cleanup and template exports. Writing scripts to harvest and parse human-written references into URIs is increasingly a process of the past, although this is still necessary for one of Nomisma's largest contributors.

More than 4000 monograms published to Nomisma

After a tremendous amount of work over the last year or more, just over 4,000 monograms that appear on Greek coinages have been published as unambiguous URIs on Nomisma.org. These monograms constitute a combination of symbols that appear on Hellenistic Royals Coinages (HRC) and Corpus Nummorum (CN). Using Computer Vision methodologies employed by Karsten Tolle, identical monograms were matched across datasets, reducing the number of monograms published by both projects from 4,500 to 4,000. The potential matches were vetted by Peter van Alfen and Ulrike Peter. The monogram URIs in HRC and Corpus Nummorum have been integrated as skos:exactMatch properties in Nomisma, making it possible to create UNION SPARQL queries that enable the querying of types (and associated Nomisma concepts) across these disparate projects. This is a huge advancement in our ability to study Greek monograms and their meaning on coinage. Even in Hellenistic Royal Coinages, we did not have the time in the first phase of the project to disambiguate identical monograms across PELLA, SCO, and PCO, which limited the data and geographic visualization of monograms within their own typological silos. 

The user interfaces in Nomisma.org have been extended to include many of the features for symbols in the Numishare platform which publishes HRC: 

• The Nomisma.org browse page can be filtered to display results only for the http://nomisma.org/symbol/ namespace, and once this filter is active, a user can refine the search results by Greek and Latin (or other) letters which appear within the monogram. It is possible to switch to a grid view that makes viewing monograms more intuitive. If constituent letters have been selected, it is possible to display a map in a popup window that contains the mints, hoards, and individual finds associated with the monogram.
  

The Nomisma.org browse page showing monograms that consist of several Greek letters.

• The page for each monogram will show associated mints, hoard, and findspots. The size of the mint points vary by the number of types that produced the monogram.
• The page will show a network graph rendered in d3 showing the relationship between that monogram and other monograms that appear on associated coin types. By default, the network graph shows the most immediate relationships, but it is possible to click another button to view the secondary relationships.

The distribution map and network graph of Monogram 1830, uniting different Hellenistic typologies.

• A list of types (and photographic examples, if available) will appear. The type list is downloadable as CSV and sortable by several categories. 
• The SVG graphics associated with each monogram are in the Public Domain and can be reused for both commercial and noncommercial publication.
  

It should be noted that although thousands of monograms from CN have been published and integrated into Nomisma, we have yet to reindex the coin type data from CN that includes their own monogram URIs. As a result, the data visualizations and type lists in Nomisma only reflect HRC. Corpus Nummorum type data will be reingested into Nomisma's SPARQL endpoint in the near future.

After publishing the RDF derived from the HRC and CN monogram data, the network graph query is formed by querying a monogram of a particular URI (or its skos:exactMatches) which appear on the obverse or reverse of a type, and extending those type queries by including other monogram URIs will appear on those same types (filtering for matches that conform to the Nomisma symbol concept scheme).

SELECT ?symbol ?altSymbol ?image ?altImage WHERE {
  BIND (<http://nomisma.org/symbol/monogram.01830> as ?symbol)
  {?side nmo:hasControlmark ?symbol}
  UNION {?side nmo:hasControlmark ?match .
    ?match ^skos:exactMatch ?symbol}
  ?type nmo:hasObverse|nmo:hasReverse ?side .
  {?side nmo:hasControlmark ?altSymbol}
  UNION {?side nmo:hasControlmark ?altMatch .
  ?altMatch ^skos:exactMatch ?altSymbol }
      FILTER (?altSymbol != ?symbol 
        && contains(str(?altSymbol), "http://nomisma.org/symbol"))  
  ?symbol crm:P165i_is_incorporated_in ?image .
  ?altSymbol crm:P165i_is_incorporated_in ?altImage .
} 

Such queries can be extended to list mints or authorities which produced them, and how many types in total are associated with those concepts. By associating types with Nomisma concepts for denominations, materials, mints, individual rulers or broader political organizations, it is possible to query and visualize relationships between entities in complex ways that simply could not be undertaken before applying Linked Open Data principles to numismatics.

Once the Corpus Nummorum data have been updated in Nomisma, I will prepare a second post discussing more complex use cases with mints and authorities.

Most of the ANS collection migrated to CollectiveAccess

Several weeks ago, the North American department was migrated from the ANS' old FileMaker curatorial database into our new CollectiveAccess database system. The intermediate clean-up work was substantial and undertaken in OpenRefine by Sami Norling, who is now a data scientist for the Smithsonian Institution.

There is substantial overlap between the persons who appear on or are responsible for the issue of coinage (including designers) between North American and Medals and Decorations, and much care was taken to disambiguate between these entities and those already created in ColletiveAccess in previous department migrations. There is also quite a large number of corporate organizations responsible for issuing tokens, notes, etc., which has gone through a rigorous normalization process to standardize names and link to Wikidata.org, when possible. The resulting quality of the North American department is substantially improved (especially when linking places to Geonames.org), which is plainly visible when using the Mantis browse page and clicking on individual object links.

To date, Sami has completed the Islamic, South Asian, and North American departments, which have all been imported into CollectiveAccess and republished to Mantis. Sami has also just completed her work on the Latin American department, which will be prepared for upload into Mantis within the next few weeks.

I completed Medals & Decorations a year ago, and Greek, Roman, and Byzantine in May, which I blogged about here and here.

Only East Asian, Medieval, and Modern European departments remain to be migrated, which will hopefully conclude by the end of this year.

Symbols and monograms on Nomisma.org

Several years ago, the Nomisma scientific committee made a decision to create a new namespace in which we would publish URIs for symbols that appear on coins (including monograms): http://nomisma.org/symbol/.

Only recently have symbols been added into this namespace, primarily religious symbols that appear on Medieval coinage. At present, these symbols link to SVG graphics hosted by the Wikimedia foundation. By the end of the year, there will likely be thousands of Greek monograms added into Nomisma, disambiguated from identical glyphs between across the Hellenistic Royal Coinages (HRC) and Corpus Nummorum typologies. This is especially important since the same monograms in PELLA, Seleucid Coins Online, and Ptolemaic Coins Online are not interlinked, even when not factoring in the integration of a project external to the HRC umbrella.

The Nomisma.org browse page now includes a new filter for Concept Scheme, enabling users to sort between only those in the /id/ namespace and the /symbol/ namespace. If the /symbol/ namespace is selected in the query, additional query options become available.

Adjacent to the pagination buttons are buttons that enable changing the view from the typical list layout to a grid, which is better for perusing the symbol graphics. Additionally, it is possible to filter monograms by constituent letters, much like in the symbol interfaces in Hellenistic Royal Coinages. A user can select one or more letters, whether Latin, Greek, or another script, to filter the monograms to those containing those letters. The perceived letters are, of course, somewhat subjective based on the specialist who undertook the work of manually itemizing them.

At present, only 2 of the 13 symbols published in Nomisma are monograms (chi-rho and tau-rho Christograms), so there is a very limited range of available letters for filtering. That will change once Greek monograms are added into the Nomisma namespace.

 

Extensions for displaying symbols in Nomisma.org browse page

Once these symbols are connected to typologies integrated as Linked Open Data into the Nomisma SPARQL endpoint, I will be able to generate lists of relevant types and maps depicting the mints, findspots, and hoards associated with them, which will create a more complete picture of their distribution over time and space, a more significant advancement in data visualization as compared to the relative silos of the HRC sub-projects.