Database Reference
In-Depth Information
lence (non-retrieved answers) in the information
retrieval phase.
As shown in Figure 1, information retrieval in
the AMI data warehouse relies on the semantic
search engine CORESE (Corby et al., 2006). The
CORESE rule language is perfectly fitted to this
kind of annotations. Indeed, CORESE provides
an inference mechanism to deduce new facts from
declared annotations. In AMI, rules are a good
mean to reflect the implicit expert knowledge.
For instance, the following information: When
the sample studied in an experimental condition
is taken from a subject affected by psoriasis,
then this condition is most likely using the IL22
inductor is an example of such knowledge that is
never explicitly stated while admitted by experts
in skin reactions. It would be translated like shown
in Figure 14.
One challenge addressed in AMI is to succeed
in building a system based on this kind of knowl-
edge rules. In a first stage, biologist background
knowledge is collected and coded by an expert.
annotations, databases and XML documents). It
is based on the semantic search engine CORESE
(Corby et al., 2006) which allows navigating and
reasoning on the whole annotation set and takes
into account the ontology concept relation hierar-
chies. New features have been added to CORESE
in order to query not only semantic annotations
but also data stored in XML documents and in
standard databases using XPATH and SQL em-
bedded in SPARQL. Indeed relational and XML
data provide relevant supplementary information
in addition to RDF annotations.
SQL Embedded in SPARQL:
An Example
In many cases, an answer to a user query should
combine annotated information with relational and
XML data. Let us consider the following query:
In which others experiments using a Nickel
treatment, CCL19 gene is up-regulated” . In this
example, the information about experiments using
a Nickel treatment can be found into annotations
on experiments, while information about gene
CCL19 behaviour (up-regulated) is stored in the
relational database.
The idea here is to query the database using
SQL embedded in SPARQL. It was introduced
Semantic Search
The AMI querying tools propose an intelligent
information retrieval system that retrieves data
stored in different types of resources (semantic
Figure 13. Example of a clustering annotation
Search WWH ::




Custom Search