Geoscience Reference
In-Depth Information
formula RSV (equation [3.2] of the previous section) in order to take into account
the roles associated with each search criterion and merge the results in a single final
list. Let us note that each mono-dimensional IRS relies on a tile-based generalized
representationofinformation.Moreover,eachIRSimplementsthevectorialIRmodel.
Figure 3.8. A meta-search engine according to Rasolofo et al. [RAS 03]
More generally, the PIV 3 meta-engine can also federate any type of search engine
and implement different result aggregation models.
In the following section, we evaluate the CMRP model implemented in the PIV 3
meta-engine.
3.5. Evaluation and discussion
As we have emphasized in Figure 2.4 in section 2.3.7, there is, to our knowledge,
no GIRS evaluation framework combining the spatial, temporal and thematic
dimensions of information. Thus, T REC [VOO 05] is a reference campaign in IR that
allows us to evaluate IRSs with respect to the thematic dimension. There are few
studies concerning the evaluation of these two other dimensions of geographic
information. The temporal dimension has been the object of the
T EMP E VAL evaluation framework [VER 09]. Furthermore, Bucher et al. [BUC 05]
have proposed the simultaneous consideration of two dimensions: spatial and
thematic. This proposition can be found in the G EO C LEF task [GEY 05] of the
C LEF framework [PET 01]. C LEF has mainly enabled the evaluation of classic
thematic IRSs in IR such as Lemur [OGI 01], Lucene [GOS 05] and
Search WWH ::




Custom Search