Thursday, October 1, 2015

On Concordances Between Type Corpora

By now some of you may have heard that the British Museum issued a take-down notice for the PELLA project, a digital type corpus of Argead coins. PELLA will ultimately generate URIs for all Argead types, and will hopefully be the canonical corpus for cataloging both museum, archaeological, and private collections. The first and most important step in building a large scale digital corpus/data aggregation system (like PELLA, OCRE, CRRO, etc.) is to build a concordance between existing type numbers and future ones. This is especially the case for PELLA, as OCRE and CRRO are merely reflections of the current editions of Roman Imperial Coinage and Roman Republican Coinage. Therefore, Martin Price's 1991 out of print, but in copyright The Coinage in the Name of Alexander the Great and Philip Arrhidaeus is just one part of the existing print corpora of Argead coinage.

The British Museum "cannot consent to the Price catalogue typology being released online until we have been able to assess whether the Trustees of the British Museum would approve the inclusion of the Price catalogue material within this online resource" because they erroneously believe that the data published through PELLA are verbatim descriptions from the corpus. We can demonstrate significant enhancements of the typology, so this brings to question whether URIs for Price types can even be minted.

The BM seemingly does not want us to issue typological data under the Price numbering system, despite the enhancements that clearly deviate from the original publication. The BM would find it impossible to enforce this demand, as Price numbers are a standard reference for the coins of Alexander the Great, and nearly every collection uses these numbers--whether commercial auction houses, large museum collections like the ANS, Berlin, or Bibliotheque nationale de France, or tiny little museums like the University of Virginia Art Museum, which owns two Price-numbered coins of Alexander. The entire purpose of numbered type corpora like RIC, RRC, Price, etc. is to create a an up-to-date system by which scholars may cite coin types in a standardized way. Price's own corpus wasn't created in a vacuum. His typologies were influenced by those of his predecessors, going back for centuries.

This brings us to the important point of concordances. While PELLA will eventually issue a new set of unique identifiers for Argead coin types (based on a numbering system combining the ruler and mint), currently, the four collections that contribute data on physical specimens to the PELLA project use Price numbers in their databases. It is impossible to aggregate these collections without first establishing a baseline numbering system. It takes little effort to map these collections to Price URIs in PELLA. Moving into the future, there will be a systematic and transparent mapping from Price to future-URIs following Linked Open Data methodologies.

The fact of the matter is will NEVER, EVER go away.

Even after we have established a new numbering system, the price.4 URI will be a semantic HTTP 303 See Other redirect to the new URI if you go to price.4 in your browser. If you request RDF for price.4 via content negotiation, it will say that this type has been dcterms:isReplacedBy ale3.amphipolis.4 (or whatever). The RDF for ale3.amphipolis.4 will say that this is a skos:exactMatch for its preceding Price number.

The Berlin Münzkabinett has already begun to incorporate Price URIs from Pella into its database. But moving forward, because of the inherent functionality in Numishare for building sustainable concordances between coin types, Berlin will be able to update their database to the most current version of PELLA type URIs through automated mechanisms. The same would be true for the British Museum, if they want in due course to benefit from the work of others freely offered, for the incorporation of PELLA (or OCRE or CRRO...or Getty or Pleiades or Wikidata, for that matter) URIs into the "5 Star" Linked "Open" Data system. This is a topic for another time, but I am feeling increasingly inclined to discuss it.

According to the British Museum, "as far as we can see this catalogue data has not been updated to incorporate any new attributions and contains no new scholarly research." This, of course, ignores the tremendous scholarly effort in mapping typologies into the graph of data in the Nomisma ontology and concept namespaces, as well as the reorganization of obverse and reverse symbols in a way that makes query possible on the Web.

This brings me to my final point:

We are also concerned that the Open Database License (ODbL v1.0) under which the PELLA project is distributed is not compatible with the Creative Commons License (CC-BY-NC-SA 4.0) license that governs all non-commercial uses of images and data available both through the British Museum Collection Online and the British Museum Semantic Web Collection Online, as well as all non-commercial uses of any documents produced by the British Museum in any format or through any medium.

The PELLA data were created by Andy Meadows and Peter van Alfen, with translations into German and French provided by Karsten Dahmen and Frédérique Duyrat, and the ANS has every right to place an Open Database License on these data, since these typologies are enhancements over what was originally provided in Price. And of course, our migration of their data from CIDOC-CRM into the Nomisma Ontology and integration of it into the Nomisma SPARQL endpoint does not violate any terms of their CC license whatsoever. PELLA is a non-commercial project. We are sharing the data alike, and we are attributing the data to the British Museum, when applicable. And we have linked to the images that they are providing through their own system, which is perfectly acceptable for their CC license. No one complained when the British Museum became partners in OCRE or CRRO, so one has to wonder why they are causing such a stink with this particular project. The British Museum is stifling scholarship and undermining the potential for numismatic research and the integration of numismatic materials into the wider cultural heritage domain. We at the ANS support Open Data and Open Access, and we hope that future scholars may make use of these data for answering more complex research questions.

No comments:

Post a Comment

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