‹header›
‹date/time›
Click to edit Master text styles
Second level
Third level
Fourth level
Fifth level
‹footer›
‹#›
SM: single operator are probably just the following: Model Generation, Mapping Composition, Mapping Alignment
Schema Reintegration -> Model Reintegration?
SM: changed “design patterns depend …” to “applications depend …”
SM: Moved from slide to comments: “Match is usually not fully automatable”. Added “Mapping reuse” instead
SM: actually, ModelGen executes a functional meta-mapping between two meta-models. Output is a target model + mapping between source and target: <target, map> = ExecuteMapping(source, meta-map);
Examples:
  execute meta-mapping (rules):<SQL, map> = ExecuteMapping(ER, rules);
  execute mapping: <XSD2, instance map> = ExecuteMapping(XSD1, xslt);
Input mappings may be partial and/or inconsistent
Reduced  complexity, no navigational code.
Beauty for programmer: Script is generic, works the other way (XML to relational, reverse engineering: rel-ER), too.
[Execution can be optimized.]
Fun part: managed to make the scenarios executable!
Intermediate results can be inspected and adjusted. Capability is important, since many tasks require human intervention: for example, FreightCharge not propagated.
Need model m to identify additions / deletions in mA and mB.
Cannot simply merge mA and mB (deletions are lost).
Map mA˘ and mB˘
Need mapping between mA˘ and mB˘
Mapping reuse pattern: line 6 and 8
mA˘ = mA − deleted by way of mB
mAx = new in mA˘
mA˘ = mA − deleted by way of mB
mAx = new in mA˘
Accent on i
grammatically, sensibly – sanctions (i.e. constraints). These terms are technical synonyms.
which, necessarily – sufficient and necessary conditions
Does the common representation allow labeled relationships? Other relationship tags (like sensibly)?
Step 3 is instead of the leaf-matching technique of Cupid, which is too expensive for a model this large.
Could cut if I’m short on time.
Say not going to talk about analysis