Lesson map
Lesson 48 published. Map, example, EX142.
flowchart TD
N0["Learning outcomes"]
N1["48.1 Contents"]
N2["48.2 Linking"]
N3["48.3 Privacy"]
N4["48.4 Integrity"]
N5["48.5 Review"]
N6["Organising the dossier"]
N7["Version control"]
N0 --> N1 --> N2 --> N3 --> N4 --> N5 --> N6 --> N7
N7 --> EX["Worked example"]
EX --> Q["Exercise EX142"]
Logical drawing
flowchart LR
N0["Learning outcomes"]
N1["48.1 Contents"]
N2["48.2 Linking"]
N3["48.3 Privacy"]
N4["48.4 Integrity"]
N5["48.5 Review"]
N6["Organising the dossier"]
N7["Version control"]
N0 --> N1 --> N2 --> N3 --> N4 --> N5 --> N6 --> N7 --> Q["EX142"]
[Learning outcomes]
|
v
[48.1 Contents]
|
v
[48.2 Linking]
|
v
[48.3 Privacy]
|
v
[48.4 Integrity]
|
v
[48.5 Review]
|
v
[Organising the dossier]
|
v
[Version control]
|
v
[Worked example] --> [Exercise EX142]
Learning outcomes
You will (1) assemble evidence for assessment; (2) link claims to evidence; (3) keep privacy.
48.1 Contents
Requirements, design records, test results, measurements, reviews, trials and operations data.
Each type of evidence supports a different kind of claim. Requirements and design records show what was intended; test results and measurements show what happened; reviews and trials show how the system held up under scrutiny and stress; operations data show how it behaves in service. Mateo includes items that show problems, such as the failed remote logging test and the trial's printer gap, because leaving them out would make the dossier less believable.
48.2 Linking
Each claim in the thesis links to dossier items.
A claim such as availability rising to 0.9999 links to the lab output, the failure records and the measured repair times, so an assessor can check it in a few steps. Claims in the thesis that have no dossier link are either supported in another way, such as a citation, or removed. Mateo keeps a list of claims and links and checks it before submission. The list is part of the dossier, so assessors can use it as a map.
48.3 Privacy
No patron data appear; operational data are aggregated.
Every item was checked for patron information before it entered the dossier. Restore logs record PCs and times, not people. Help-desk data were stripped of names before analysis and are summarised as counts by category. Comment cards are quoted only where the patron could not be identified. Fernhill's data protection officer reviewed the index and agreed that nothing in it identifies a patron. Her review is recorded in the dossier with the date and the version of the index she checked.
48.4 Integrity
Files carry digests so assessors can confirm they are unchanged.
Each digest is computed when the item is added and listed in the index. An assessor can recompute the digest of any file and compare it with the index, which shows the file has not been changed since submission. Mateo uses the same hashing approach as the manifest work earlier in the course. Corrections after submission are added as new versions with their own digests, so the history stays visible.
48.5 Review
Dr Novak checks a sample of links.
Dr Novak picks twenty claims at random and follows each link to its evidence. Two links point to the wrong version of a file, which Mateo corrects before submission. Sampling like this cannot prove every link is right, but it shows whether the index is reliable enough for assessors to trust. Her checklist and results are stored with the dossier. Mateo then checks every remaining link himself, since the sample found errors, and records the result.
Organising the dossier
The dossier is organised by requirement and by chapter of the thesis. An index lists each item with a short description, date and digest. Assessors can enter through a claim in the thesis or through a requirement. Mateo tests the index by asking a classmate to find five items; all are found within minutes.
Large files such as logs are summarised, with full versions available on request.
Version control
The dossier is versioned, so later corrections are visible.
Retention
Fernhill keeps a copy for its records.
Access
Access is limited to assessors and Fernhill IT.
Final check
Before submission, every link is tested.
Archive
The dossier is archived with digests.
Assessor guide
A one-page guide explains how to navigate it.
Completeness
Every requirement has at least one evidence item.
Readability
Summaries use plain language.
Updates
Late evidence is added with dates.
Structure example
For requirement R4, the dossier links the requirement record, the restore control specification, the lab output from chapter 28, the timing data from chapter 29, the 30-night acceptance record and the trial report.
Assessor feedback
Assessors note the dossier is easy to navigate and complete.
Digests
Each item carries a digest so assessors can confirm it has not changed since submission.
Worked example
An assessor picks the availability claim and follows links to the lab output and failure records in under five minutes.
Synthetic case. Synthetic.
Errors and limits
Claims without evidence weaken the thesis.
Exercises
- EX142 — Contents. List four dossier items.
Show marking criteria
Rubric: 3 = requirements, designs, tests, measurements, reviews, trials, ops; 2 = four; 1 = two; 0 = none. - EX143 — Privacy. How is privacy protected?
Show marking criteria
Rubric: 3 = no patron data, aggregated; 2 = partial; 1 = vague; 0 = none. - EX144 — Integrity. How can assessors confirm files are unchanged?
Show marking criteria
Rubric: 3 = digests; 2 = partial; 1 = vague; 0 = none.
Synthesis
A linked dossier makes claims checkable.
Continue studying — chapter 48 of 50
← Chapter 47 (draft) · Course chapter index · Chapter 49 (draft) →