Thanks to the Met Museum‘s Open Access policy, we have been able to integrate more than 14,000 objects from the museum’s collection into Kerameikos.org. These objects were extracted from the Met’s CSV export to Github, supplemented with image URLs from a derivative repository. This tranche of data does not include all ancient Greek and Greek-aligned (e.g., Cypriot) ceramics, but only those filtered from approximately 1000-400 BCE. Therefore, the dataset includes Geometric pottery and the vast swath of Archaic and Classical objects, but ends before the Hellenistic period. The majority of these objects are fragments, but there are many intact objects among the collection, which are accessible by visiting related Kerameikos concepts. For example, the integration of the Met’s collection into the Greek pottery Linked Open Data cloud has added three additional vases to the example specimens of the Archaic potter/painter, Exekias.

There are a few key pieces of data missing from this objects, however: 1) Only the main image is made available in Kerameikos, from the supplementary CSV file from Github since there are API rate restrictions that prohibit us from extracting more URLs for display, 2) The CSV and API do not include the technique (e.g., Black-figure), even though the technique is visible on the object’s web page on the Met’s collection database site, and 3) the provenance (findspot) is also not included in the public data, although it must certainly be recorded internally. That makes these data somewhat limiting for geographic query and visualization.

Nevertheless, this is the largest dataset integrated into Kerameikos so far since sherds are included, although they had been mainly excluded from the British Museum material we harvested. We do have flags in our source OpenRefine data regarding whether an object is fragmentary or not, but have not introduced the state of conservation into the CIDOC-CRM linked data in Kerameikos (which conforms to the Linked Art profile), which would enable the filtering of complete and incomplete objects in the user interfaces. We will likely implement this, eventually.