The work aims at developing an approach for flexible and efficient
answer to data mining queries addressed either to a relational or a multidimensional
database. The authors investigate the two following techniques. The first one exploits
lattice based mining algorithms for generating frequent closed itemsets from
multidimensional data. The second technique uses new operators, similar in spirit
to the OLAP ones, in order to support data mining on demand. These new operators
working on concept lattices include projection, selection, and assembly.
Chapter XII, Temporal Semistructured Data Models and Data Warehouses, by
Carlo Combi and Barbara Oliboni, extends the application of the DW technology
to semistructured data that may evolve in time. In order to store and analyze data of
this kind, the authors propose a graph-based temporal semistructured data model.
In this model, semistructured data are represented as a labeled graph with complex
nodes representing abstract entities, simple nodes representing primitive values as
well as with edges connecting nodes. Labels are associated with nodes and edges.
Temporal aspects of data are handled by including valid time intervals in labels.
In order to assure the consistency of this model, the authors propose two kinds of
integrity constraints, namely, basic and domain-dependent ones. Basic constraints
have to be satisfied by every graph, whereas domain-dependent constraints can
be defined either for some specific nodes and edges or for the whole graph for a
specific application domain.
Pages:
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40