Friday, February 25, 2011

BiSciCol Technical Meeting: University of Florida, 23-24 February 2011

The BiSciCol Technical team had a successful brainstorming meeting over two warm Florida days.

Nico Cellinese (UF), Kate Rachwal (UF), Reed Beaman (UF), Gustav Paulay (UF), Russ Watkins (UF), John Deck (Berkeley), Brian Stucky (Colorado University), Rich Pyle (remotely, Bishop Museum) and our very special guest Steve Baskauf (Vanderbilt University).

The main goal of the meeting was to define the BiSciCol scope, techcnical goals and the model. Happy to report we made great progress. We are now ready to move to an implementation phase and are confident will be able to release a prototype in July 2011. See below a snapshot of out design document.

Purpose of the System
1. Notify all objects that are related to Object X that Object X has changed.
2. Manage relationship definitions between objects that exist in distributed databases.
3. Manage annotations of objects where annotation is treated as an Object relationship.

Technical Goals of the System
1. scalable design
2. easy coding but must be robust
3. expandable to different domains

Envisioning this as a distributed network with structured content defined. Goal is to allow for multiple data sources, each with their own thematic content. The thematic content is built on defining relationships between objects.

The Model
Linkages between objects is unstructured which allows any object to be tied to any other object via the predicate “relatedTo”. This is similar to SKOS “related” but we have made relatedTo transitive. relatedTo does not require objects to be organized hierarchically.

Following are the triples recognized in the network:

:objectId :relatedTo :objectId
:objectId rdf:type :objectType

:relatedTo :hasProperty :propertyDefinitionURI
:objectId dwc:dateLastModified “YYYY-MM-DD”

Monday, February 14, 2011

News and updates

Since our last meeting in Washington, DC we worked out some of the object relationships and how we want to formalize concept/objects such as individuals, specimens, vouchers etc. for the purposes of our network. We also agreed to consider implementing RDF in our design. To this end, a number of people have since purchased and read (hopefully past the introduction) the book "Semantic Web for the Working Ontologist: Effective Modeling in RDFS and OWL". Finally, we agreed to deliver a working prototype of the BiSciCol network in July 2011, implementing data from Biocode, CalPhotos, and UF.

During January and the first part of February, John Deck (Berkeley), Rob Guralnick (Colorado U.), and Brian Stuckey (Colorado U.) have been crafting a technical implementation plan for the BiSciCol prototype. CU has offered a server and Tomcat instance for our prototypes and Brian has been implementing that along with getting the web interface going. John has been working on a model that ties together collecting events, specimens, tissues, DNA extractions, and photographs (more object types later), while incorporating location and modification date. Taxonomy is notably absent for now and we will soon be working more closely with Rich Pyle and Rob Whitton (Bishop Museum). Codebase is Java and inferencing/RDF work being implemented in Jena/ARQ (open source Java).

On the 23rd and 24th of February, John and Brian will be at UF for a technical meeting with Nico Cellinese, Kate Rachwal, Russ Watkins, and special guest Steve Bauskauf (Vanderbilt University) to review the UF implementation and proposed model. We will also work remotely with Rich Pyle and Rob Whitton. More details on this meeting will be posted soon. Given our prototype deadline of July 2011 we have a lot of ahead of us, especially in integrating taxonomic names components but we feel we are on target with progress to date.