Semantic resources project/Semantic Resources Architecture

JAR has written a document describing the Neurocommons architecture.

Alan's SWAT 2009 Presentation contains a reasonable diagram of the Neurocommons architecture (cf pg. 37)

Architecture Description
Our architecture contains two distinct sections, joined through a set of common interfaces and data formats.

The user-facing part of the architecture consists of the Science Collaboration Framework (SCF), a tightly-integrated collection of web software components for building online scientific communities. SCF is implemented using the Drupal Content Management System (CMS), and is itself the basis for custom modules developed to support individual scientific communities. In our architecture, each SCF installation is accompanied by a semantic data source, a "triple store," which holds the community-generated or community-required semantic data relevant to that SCF site.

These semantic data stores are populated from, and synchronize to, the second half of our architecture: the Neurocommons triple store and data import process. Neurocommons is the name for a data import and storage pipeline which begins with the web itself, and a manual data editing process. Data is curated and stored in an interlinked, semantic format as distinct modules, called 'bundles.' Bundles are then exported, individually or in groups, to downstream users.

Interface through SPARQL and SPARUL, timed (batch) uploads and downloads, common data formats and packaging from SCF back to Neurocommons?

Neurocommons
The Neurocommons infrastructure depends on two essential units of software: the open-source Virtuoso triple store, distributed by OpenLink Software Corporation, and an internally-developed software package for the marshaling, export, and import of RDF bundles known as 'rdfherd'. We have developed a custom version of the Virtuoso server software to support the specific needs of bundle storage in this project; this development has included building relationships with the OpenLink Software Corporation developers and contributing software patches to their open source development team. The second tool, 'rdfherd', is a software package developed in the course of this project specifically intended to support the import and manipulation of data "bundles" such as the resources we are developing. These tools are used to support both regular imports from third-party data servers (such as the OBO Foundry), as well as to provide a publicly-accessible query access point for all the Neurocommons resources.

Revised Architecture Diagram
The [[Media:SemanticResourcesArchitecture.graffle.gz | gzip'ed graffle]] of the very rough draft of an architecture figure for the Semantic Resources Project.