Database Reference
In-Depth Information
Figure 7.1 The process of extracting knowledge from the data using the progressive mining
technique, restricting the constraint at each level.
Model Manipulation
Similarly to the trajectory data, the models resulting from the mining step can
be stored and manipulated to produce a useful and meaningful representation
of the trajectories behaviors. For this reason, the
relation statements
and the
transformation statements
can be used also on models. In particular, M-Atlas
provides a relation that is a bridge between data and models called
entails
,
which identifies the data that support a model, realized with the following
query:
CREATE RELATION <relation_table> USING ENTAILS
FROM (SELECT t.id, t.object, m.id, m.object
FROM <trajectories_table> t, <models_table> m)
Notice the use of
ENTAILS
keyword in the query. The idea is to apply the
entails operation to the join between trajectories and extracted models specified
in the
SELECT
statement. This relation is crucial to the knowledge discovery
process, as it implements the interaction of the process building complex pro-
gressive queries between data and models. This procedure is called
progressive
mining
and it is illustrated in the following paragraph.
Progressive Mining
As described in previous sections, the knowledge discovery process is not a
straightforward sequence where a single run of data mining algorithm can per-
form the whole understanding task. The iterative and interactive aspects are
crucial to get a real understanding of the data and extracted patterns. The pro-
gressive mining technique is the concatenation of a series of mining algorithms,
which restrict, at each step, their constraints, removing the
not interesting data
or
noise
. Figure
7.1
shows a graphical representation of the process where at
each step the models are extracted and the data supporting them are reused
