Sample Ad Advertise your business on myplick. Only $2.00 a month.
Comments:
Notes:
Slide 1: Model Fusion ≡ AOP ≡ Feature Interaction
Tom Maibaum McMaster University tom@maibaum.org
Slide 2: Some History
Views idea (models in different formalisms used jointly to design systems) originated with:
Atendolfo Angel Pereda Borquez and Roberto Lins de Carvalho, PUC/Rio, 1978-82 mapping models from one formalism to another used 4 different formalisms Celso Niskier, Daniel Schwabe and Carlos José Pereira Lucena, PUC/Rio, 1984-86 built tool PRISMA to represent in Prolog cluses the information from 3 models, to maintain joint consistency Celso came to Imperial College to do a PhD, introduced the ideas to Anthony Finkelstein and PRESTO Bashar Nuseibeh, Steve Easterbrook, etc
September 13, 2007
CASCON:Model Fusion
2
Slide 3: Related Ideas
Database integration and fusion:
originally, mapping various DB schemas into some ‘fusion’, so as to integrate respective DBs with modern genetics DBs, with badly defined schemas and humongous amounts of data, emphasis on logical (as opposed to real) integration the idea is to provide a mechanism for joint use of DBs for computational purposes, without actually building a joint model
September 13, 2007
CASCON:Model Fusion
3
Slide 4: Relating Models
Underlying any notion of views/model fusion is a map from one formalism to another
how to map a model in one into a model in the other integrate a view into a global model relate one view to another
This is straightforward enough, BUT What happens when structure is taken into account?
September 13, 2007
CASCON:Model Fusion
4
Slide 5: Relating Models
Different formalisms with different concepts of object (component) and morphism (structuring mechanism)
September 13, 2007
CASCON:Model Fusion
5
Slide 6: Relating Models
… if formalisms closely related; structural elements are inductively mapped to compound structures in other formalism
September 13, 2007 CASCON:Model Fusion 6
Slide 7: Relating Models
… if formalisms not closely related; structural elements are smeared* across the whole structure in other formalism
* see any decent category theory book for a definition
September 13, 2007 CASCON:Model Fusion 7
Slide 8: model fusion results in smears on smears*
* see any decent category theory book for a definition of smear composition
September 13, 2007 CASCON:Model Fusion 8
Slide 9: Summary
This puts us in mind of aspect oriented programming … which puts us in mind of what was the feature interaction problem of telephony
fused structures lose structure a lot of detail related to simply representing structure indirectly feature interaction occurs order of fusion becomes important
WHAT A MESS!!!!!!!!!! If you don’t want to blow a fuse, don’t literally fuse!
September 13, 2007 CASCON:Model Fusion 9