This work was performed as part of the project to pilot linked data conversion, publication, and visualization of Harvard Geospatial Library metadata and Harvard Film Archive metadata.
The HFA (Harvard Film Archive) project created linked data descriptions for a set of moving image materials by women directors--work that has previously been underexposed and in many cases is unique to the HFA. The overall HFA project is described at Harvard Film Archive
See Harvard LD4L Labs wiki for documents
HFA subjects & genre mapping to LCSH, Getty AAT, and FAST URIs
Converted a full snapshot of the Harvard Film Archive metadata to the target Moving Image linked data ontology (https://github.com/HLITS/LD4L_Film_Ontology )
Vitrolib custom form for annotations
Vitrolib lookup specs for ISNI
Discussions with Library of Congress BIBFRAME pilot participants
Pattern documents for LD4P/LD4L Labs BIBFRAME extension group
LD4P/LD4L ontology extension meeting
Interviews with Harvard Film Archive and Northeast Historic Film staff
HFA data originated in FilemakerPro format. A java program was created using FilemakerPro database drivers to extract data from two relevant database tables. This data was output to XML format in a large single file for processing by the BIBFRAME converter described below.
Harvard created native Linked Data descriptions for a selection of library cartographic resources including printed maps, atlases, digital geospatial datasets, and other cartographic information resources. Together with LD4L-Labs partners, Harvard specifically converted a set of Harvard Geospatial Library metadata records into linked data descriptions. The overall Harvard Cartogaphic Materials project is described at Harvard Cartographic Materials
FGDC metadata from Harvard Geospatial Library originated in XML format so no converting of this format was necessary prior to processing by the BIBFRAME converter described below.
The bib2lod project (see also MARC -> BIBFRAME Converter Framework) was used as base code for converting both the HFA and FGDC XML data. An extension of this base code was made for each of these input formats. Custom code for each project was necessary due to significant difference between the datapoints available for each format. The XML input for each of these formats was converted to RDF output. This RDF output was imported into a Vitrolib web application, one for each format type. During the development process extensive test cases were written for each format type and vetted a domain expert.
https://github.com/ld4l-labs/fgdc2lod
https://github.com/ld4l-labs/hfa2lod