Module 02 Activity

Scenario

You have the corpus from Module 1. Now it has to be divided into passages, and the decisions here determine what every citation in the system will look like.

What you produce

A chunking and metadata specification, plus a hand audit of thirty passages against the stand-alone test.

Steps

1. Choose the semantic unit and write it in one sentence - one policy statement, one procedure step, one definition. Justify it against the citation test in Unit 00.
2. Mark up three documents by hand at that grain. Count the passages and note the shortest and longest.
3. Specify the five metadata fields from Unit 02. For visibility, say where the value would come from and who owns it.
4. Apply the three stand-alone checks from Unit 04 to thirty passages by hand. Report the failure rate, not just the failures.
5. Re-mark the ten worst and report the rate before and after.
6. Write the citation id format and show what a reader sees when they follow one.

Evidence to hand in

• The chunking rule in one sentence, with its justification.
• Thirty marked-up passages with all five metadata fields.
• Failure rate before and after re-marking, with three examples of what was wrong.
• The citation id format with one worked example.

Review checklist

• Passage size follows from the semantic unit, not from a length target.
• Every passage has a visibility value with a stated source and owner.
• The stand-alone failure rate is reported as a rate.
• At least one passage that passes the mechanical checks is shown to be unusable anyway, or the audit says none was found.
• The id format would survive the corpus being reorganised.

Mark complete