Chapter 5 Repair Protocol
Use this after actual performance data exists. Do not pre-label mastered/weak areas based on guesswork.
1. Key terminology is weak
Redraw the construction-vs-function key taxonomy. Then classify at least five keys that have multiple simultaneous labels.
Required distinctions: - simple/composite/compound; - super/candidate/primary/alternate; - business/surrogate; - identifying/non-identifying migration.
2. Scheme questions are weak
Rebuild the six-scheme map and say the dominant problem each helps represent:
relational / dimensional / OO / fact-based / time-based / NoSQL.
Then classify four NoSQL storage shapes.
3. Model-level questions are weak
Create three columns: CDM / LDM / PDM. Sort at least 20 details into them, then add a fourth bucket for canonical data-in-motion structures.
4. Normalization is weak
Rebuild 1NF/2NF/3NF and create one original defect each removes. Only after that, add recognition notes for BCNF/4NF/5NF.
5. Activity/process questions are weak
Recite forward and reverse sequences, then rebuild the conceptual/logical/physical build steps and explain why iteration can move backward.
6. Governance questions are weak
Rebuild:
Standards → Design Review → Version/Change Control → Scorecard → Maintenance
Then reconstruct all ten Scorecard categories and explain at least three changed-fact cases that shift the failing category.
Error-recording rule
For each miss, capture: - question ID; - your answer; - correct answer; - confidence; - time; - confusion pair / reasoning error; - repair exercise completed; - retest date/result.
That error history will later determine Artifact 08.