Abstract
The availability of automatic tools for inferring semantics of database
schemes is useful to solve several database design problems such
as that of obtaining cooperative information systems or data warehouses
from large sets of data sources. In this context, a main problem
is to single out similarities or dissimilarities among scheme objects
(interscheme properties). This paper presents graph-based techniques
for a uniform derivation of interscheme properties including synonymies,
homonymies, type conflicts, and subscheme similarities. These techniques
are characterized by a common core: the computation of maximum weight
matchings on some bipartite weighted graphs derived using a suitable
metrics to measure semantic closeness of objects. The techniques
have been implemented in a system prototype. Several experiments
conducted with it, and (in part) accounted for in the paper, confirmed
the effectiveness of our approach.
- bipartite
- closeness,
- conflicts
- cooperative
- data
- database
- databases,
- distributed
- graph-based
- graphs,
- heterogeneous
- homonymies,
- information
- matchings,
- maximum
- metrics,
- object
- schemes,
- semantic
- semantics,
- similarities,
- sources,
- subscheme
- synonymies,
- systems,
- techniques,
- type
- warehouses,
- weight
- weighted
Users
Please
log in to take part in the discussion (add own reviews or comments).