The Linked Data for Libraries (LD4L) project aims to create a Linked Open Data (LOD) model that works both within individual institutions and across libraries to capture and leverage the intellectual value that librarians and other domain experts add to information resources when they describe, annotate, organize, and use those resources. First we developed a set of use cases illustrating the benefits of LOD in a library context. These served as a reference for the development of an LD4L ontology which includes bibliographic, person, curation, and usage information. This largely draws from existing ontologies, including the evolving BIBFRAME ontology. We have prioritized the ability to identify entities within library metadata records, reducing reliance on lexical forms of identity. Whenever possible we seek out persistent global identifiers for the entities being represented — identifiers from established efforts such as ORCID, VIAF, and ISNI for people, and OCLC identifiers for works for example. One group of LD4L use cases explores circulation and other usage data as sources that could improve discovery, and inform collection building. We are exploring the use of a anonymized and normalized metric that may be shared and compared across institutions. Ontology work and software from the LD4L project is available from our Github repository. SWIB15 Conference, 23 - 25 November 2015, Hamburg, Germany. http://swib.org/swib15 #swib15