Mapping-Based Merging of Schemas - Schema Matching and Mapping - page 232

Map A_B

Model A Model B

m 1

Actor Actor

Actor
Name

All
Bios

First
Name

Last
Name

Actor
ID

m 3 m 4

ActID Bio Bio

m 5 m 6

m 7
Name =
“Official”

m 8
Name=
“Unofficial”

Fig. 8.4 An example mapping for the merge in Pottinger and Bernstein [ 2003 ]

Actor

ActorID AllBios

ActorName

Official Unofficial FirstName LastName

Fig. 8.5 The result of performing the merge in Pottinger and Bernstein [ 2003 ] on the models and
mappings in Fig. 8.4

result of performing the merge in Fig. 8.4 . The diamond-headed intra-model edges
indicate a containment relationship. The double arrowed inter-model relationships
indicate equality between elements. Looking carefully at Map AB reveals that there
is an element All Bios , which is not mapped to by any element, but contains the
Bios from models A and B , both of which have been renamed.
This complex mapping allows users to generically yet accurately represent the
way that the elements are related to each other. As with other mappings, creating
this style of mapping is beyond the scope of this paper; [ Wang and Pottinger 2008 ]
studies how to create these mappings.
Vanilla Merge defines what the output should be based on the principles laid
out in many other papers referenced here, including the least upper bound notion
from Buneman et al. [ 1992 ].
While Vanilla Merge defines that there may be other fundamental conflicts, the
fundamental conflicts in the Vanilla meta-meta conflict are the one type conflict
(an adjustment of the conflict discussed in Buneman et al. [ 1992 ] for the Vanilla
meta-meta-model) and acyclicity.

6.3 Providing Choices to Users

Chiticariu et al. [ 2008 ] concentrate more on how the user interacts with the system.
In particular, the authors assume that they are trying to merge a set of schemas. They
model these schemas as graphs containing elements related by has-a relationships.