Extending The OpenCitations Codebase For Creating Non-DOI Citation Indexes: The Ingestion Of The NIH Open Citation Collection

This thesis introduces the OpenCitations Index codebase expansion implemented for the creation of NOCI, a new citation index based on the National Institute of Health Open Citation Collection (NIH-OCC). Since the entities in NIH-OCC are identified with PMIDs, the software infrastructure was adjusted to include and manage the new identifier typology into a DOI-based system. The adjustments allow the possibility of handling duplicated data, i.e., adding the same citation twice (both as a DOI-to-DOI and PMID-to-PMID form) into the OpenCitations dataset. To handle this issue another type of identifier for bibliographic entities was adopted, i.e., the METAID, to unambiguously identify a bibliographic entity in OpenCitations, tracing back all its identifiers to the same METAID. Both the data harvested for populating NOCI and the mapping information generated are stored as RDF triples and made available in a triplestore, queryable using SPARQL language.Simplified data-access alternatives are provided, such as an API service and data dumps. The development of the software expansion required programming in Python and using the Semantic Web technologies, and a theoretical background on the OpenCitations Data Model. The achieved software flexibility will be beneficial for OpenCitations not only for the NOCI integration, but also for future ingestions dealing with PMID and other types of identifiers. This thesis starts by reviewing the citation data and how these are stored and represented in OpenCitations, and moves toward the definition of a methodology for the creation of NOCI. This approach involves the development of software expansions on the current OpenCitations software infrastructure, which is discussed in a both theoretical and technical perspective. Finally, the new software is evaluated through an analysis of its computational costs