Monday, August 3, 2020

First steps made in the normalization of the ANS Islamic department

Following some recent upgrades to our data consistency in MANTIS in the linking of many Greek names to Nomisma URIs, I have run a first-pass cleanup of entities that appear in the authority fields in the ANS' internal curatorial FileMaker Pro database. As part of the Egyptian National Library project, a large number of Islamic rulers, dynasties, and corporate bodies were created to cover the range of entities in the ENL's collection. This work was largely undertaken by Jere Bacharach.

Recently, I attempted to run the ANS's list of names that appear in the Authority facet in MANTIS against the OpenRefine reconciliation API. I have successfully matched more than 300 labels within our database to Nomisma URIs. In some cases, a single entity was represented by more than one string in our database, and so we have made some noticeable improvements in consistency. We have also filtered out any value from FileMaker that is obviously not useful. For example, several hundred three-digit number codes had been inputted. Current curatorial staff is not even aware of what these codes correspond to, and so there's really no point in publishing these values into MANTIS.

1930.168.89, an Ilkhanid coin

1930.168.89 in Arabic

Nearly all of the Nomisma URIs for people link to dynasties or corporate entities, which have been indexed into MANTIS in a "State" facet. We plan to review the relationships between these entities, as dynasties and corporate bodies have been conflated in Nomisma, and we should split these into two separate classes of concept. For example, the "Abbasids" ( carry a definition that corresponds more to a corporate entity, e.g., the Abbasid Caliphate ( An Abbasid Dynasty entity should also be created in Nomisma to connect rulers together by means of familial relationships, but individual governors operate under the authority of the Caliphate without being relatives in the Dynasty. An additional benefit of these entities being connected to Nomisma URIs is that translations for all of them are available in Arabic, among other languages.

The cleanup is done in an independent PHP script that processes CSV exported from FileMaker into NUDS XML for publication into Numishare. An intermediate lookup table is implemented by matching a value the relevant field in FileMaker to a key in a Google spreadsheet in order to extract a Nomisma URI in another column, if available. About 300 of 2,500 total lines have been matched to URIs. But this will be a jumping off point for doing more comprehensive cleanup and Nomisma ID creation later, as we migrate into CollectiveAccess.

No comments:

Post a Comment

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