A Model for Using DataCite DOIs in Observatory Bibliographies

The Chandra Data Archive's bibliographic database has been using IVOA persistent identifiers (PID), issued under the authority of the ADS, to link its datasets to articles in the literature. This type of identifier was established in 2002 by the NASA astrophysics data centers and has served us well, as it had enabled us to provide complete literature-data linking at a fine-grained level, resulting in the most complete observatory bibliography currently in existence. However, at this point it is time to transition to international standard identifiers that have come into existence after we started developing our bibliography. The obvious choice is the DataCite DOI. However, it is not an entirely trivial conversion. Just like each publication has its own unique PID, it is essential that each dataset receive its unique PID. The problem that arises is the aggregation of PIDs for any particular article. Currently, the ADS performs the aggregation function for the IVOA-type identifiers, but that will not be feasible for the DataCite DOIs. Aggregation is necessary, since dataset PIDs need to refer to a landing page and separate landing pages for each dataset would be extremely user-unfriendly. The solution, adopted by some, to mint a single PID for each article that holds all the datasets is not acceptable, since it violates the requirement that each dataset have a unique PID. The solution is to mint two types of PIDs, one for articles, the other for datasets, that refer to each other using standard DataCite DOI metadata elements. A related initiative is the Scholix project, under the auspices of the Research Data Alliance (RDA).