We developed CERIF-2000 ontology expressed in DAML + OIL (http://www.daml.org) language for research data.
DAML + Oil Ontology for CERIF + DublinCore and Math-Net metadata set
The ontology is vocabulary to describe such objects as projects, persons, organizations, results, publications, equipment, their attributes and relations between them. Having vocabulary, data can be described in RDF with defined semantic and we can use that data to answer for users' requests, load into database, etc
Show case step-by-step description
Step 1. The researcher uses
OntoMat and
CERIF-2000 ontology to describe data about
himself and his projects, publications.
Researcher opens OntoMat, then load ontology which specify which entities, which
attributes and relations he can describe. Then using own html pages (example:
http://derpi.tuwien.ac.at/~andrei), drag
and drop capabilities of OntoMat and inputting data manually, the researchers
creates metadata describing research. To create metadata, for each entity
one creates object of known class (hierarchy of classes from ontology is
visualized by OntoMat), and fill in attribute values, then defines relations
with other objects.
The researcher save new page with metadata embedded into the disk from OntoMat (OntoMat
can savenew page which is copy of loaded with embedded RDF metadata) . And then
copy it into public-accessible web site
Step 2. To make data known for AURIS-MM, the researcher register page in
the AURIS-MM Research Agent facility. One uses page/site registration form to do
ir. Researcher only registers url of the html or rdf page, or
provide additional descriptions to determine context of data.
Step 3 AURIS-MM agent (now based on
RDF Crawler)
visits page and and pages referenced by it, get RDF metadata, load
them into AURIS-MM RDF datastore. So, finally one datastore (currently file) is
created which contains all metadata collected by agent.
Step 4. Information seeker use AURIS-MM Query interface to find data
about
researcher or project/publication. Searching information, he loads ontology
which provides vocabularies for search forms and search interfaces. Search
interface understand class relations (in future other semantic expressing
a.) in search. AURIS-MM search agent use
Jena toolkit to parse RDF, DAML
ontologies, investigate semantic relations and query data (RDQL)
Also very little report capabilities should be demonstrated (like ...
how
many report in such area in a last month)
5 Someone annotated research data/page with additional information. It
might
be done in AURIS-MM directly or in tools like Amaya
6 One register his nnotation in AURIS-MM
7 Information seeker when find data can ask for annotations
Information seeker cans earch annotation or use them in information
search
Show case 2
The same as 1-4 of previous but
University has own vocabulary for publications. If example, AURIS-MM use
CERIF-2000 (based on MARC), but university has also "Professor work" type
of
publications
University would like to publish that information into AURIS-MM but
without
lossing of information
So University pubslishes also some DAML statements describing that
professor
degree is a subtype of ... (we should define what is it in CERIF) maybe
dissertation
Every user which do not know about this Austrian specific publication can
search this publications as a dissertation
but on his request he can get more detailed description of the publication
type
Used technologies and tools
Developer
OntoEdit, Protege-2000, OilEd, FaCT as ontology development tools and
maintaners (should be finally choosed the best set)
RDF Crawler to collect published RDF and create database
HPL Jena as a parser and query facilit for RDF (and Sesame?)
Might be some Prolog enhancement for querying
User
OntoMat to annotate data about research
AURIS-MM registration form
AURIS-MM annotation page to annotate others research
AURIS-MM search form to search information
AURIS-MM report form to get simple reports