Semantic resources project/Experiences

As we proceed with the Semantic Resources Project, we will try to document some of our experiences, both in social and in engineering terms, as a reference for researchers and curators in the future. An example of such an "experiences" document, which we have found useful in our work for this project, is the "Experiences with the conversion of SenseLab databases to RDF/OWL" document produced from the Neurocommons import of the SenseLab dataset.

Cooperation with existing ontology efforts has been an important part of our planning and execution of this project. In particular, we are designing our antibody resource so that our modeling efforts mesh well with the ongoing protein modeling (PRO) initiative. The PRO ontology effort includes a web interface for the human submission of new term requests; our initial antibody dataset contains thousands of references to antigenic proteins which must have terms assigned to them, too many to enter by hand. Our goal has been to build, under guidance from the PRO team, a set of technical interfaces between a large modeling project and the PRO ontology, so that requests can be submitted and processed "in batch." Initially, this technical interface is a set of file formats reflecting the submission standards to PRO. Eventually we hope to include software tools for the assembly of batch submissions, which may be of interest to other large-scale modeling or research efforts that need to interact with PRO.

A second aspect to our work has been collaboration with the AlzForum editors, working biologists who are involved in the active curation of knowledge about Alzheimer's and neurodegenerative disease into semantic elements for the SWAN Knowledge Base. AlzForum has its own independent editorial process which acts as a gatekeeper for content on its site and in SWAN. As part of our integration work with SWAN, we have interviewed the AlzForum editors in order to understand their curation requirements. We need to be able to carefully track the provenance of both the data that we bring to the SWAN effort (for gatekeeping purposes) as well as the assertions that we import from SWAN into other resources (for correct attribution).