SWAN presentation to MIT Semantic Web meetup group by Paolo

Our goal: Manage the [scientific knowledge base]

hypothesis
contains 40 statements
including: A claim
which has:
qualifiers
supporting evidence
relationships with other statments in the knowledge base

not a complete RDF graph because we are pruning some information
first level, claims (C)
then in the outside level
H stands for hypothesis

The Golde hypothesis includes a claim which is inconsistent with another claim in the same hypothesis. This suggests something is wrong here.

Curation process: in this case, the authors of the journal article have been working with our SWAN curator, Gwen Wong, PhD.

All the knowledge is represented in the

SWAN Ontology
taking into consideration SCF

Covers the SWAN Project (and

Considers the integration of domain ontologies provided by other communities (GO, bibo, PRO...)

It is pretty general, and designed or re-use (full or partial) in other scientific contexts

it is in fact a collection o

We have to glue together a lot of other ontologies.

Our core business of course is the Scientific Discourse

a few classes

relationships changed
relatedTo (5) - refersTo (4) - inResponseTo (3) -disagreesWith, agreesWith, discusses
(and all the rest)

working with DERI for integration with SIOC

1.2 not officially released yet. Still working with provenance, authoring, and versioning. Movig back to Dublin Core for more.

Agents, FOAF, Collections, Citations as well as Provenance, authoring, and versioning

JournalArticle

Citations Provenance
Gene Record
Life Science Entities (inc GeneRecord)

we like to keep relationships between real things.

Ongoing relationship with Science Commons, it may change.

Last version of the ontology has
controversial, some people don't like it, others like it a lot
People managing communities like we do like it.

These controlled vocabularies come from scientists.

Sometimes they create ontologies
not perfect, collect heterogeneous things together
that's fine, we want to add these taxonomies to our system simply, and maybe later map

So we define with SKOS
and have a set of properties we use for mapping
this method introduces one level more between ontologies and [closed] vocabularies
we won't try to

been working to integrate SWAN with SIOC

loose relationships

When you guys started using the new version, how did adapting the SWAN application to the new version work?

It is a Java application, every time we update it it is pretty hard

JSON serialized from RDF
pretty expensive

Moving to Drupal
[and i didn't hear how this will make it better]

Resolution