1. Why a semantic layer | Semantic Layer course

Section 1

The Tuesday meeting

Next section

Section 2

Where the meaning was hiding

Next section

Section 3

What a semantic layer actually is

Next section

Section 4

Why agents raise the stakes

Next section

Section 5

ACE: the three things you are buying

Next section

Section 6

What you will build

Hands on

Reproduce the disagreement

15 min

Run three defensible SQL queries against one loan book and get three different default counts.

make lab-up            # once, starts Postgres + Fuseki + Ontop
make lab lab01-two-answers

Full instructions, starter files and the solution are in labs/modules/lab01-two-answers/README.md. The lab checks its own work, so you will know if it passed.

Check yourself

Quiz

6 questions, pass mark 70%. Answers are explained either way, so a wrong guess is not wasted.

1 Three teams report different counts for loans in default, and no team made an arithmetic error. What is the cause?

The database has corrupted rows Each team encoded a different definition of default in its own SQL One team queried a stale replica The loan table needs an index on dpd

2 A semantic layer is best described as:

A cache that speeds up warehouse queries A shared, machine-readable description of what the business means, sitting between raw data and whoever is asking A replacement for the data warehouse A dashboard product

3 Which two things go wrong when an agent queries loan_master with no semantic layer? (pick two)

Select all that apply.

It has to guess which of four columns means the loan went bad It cannot generate SQL at all It cannot show why it chose that definition, so the decision is not auditable It exhausts the database connection pool

4 ACE stands for:

Agentic Compute Engine Accuracy, Consistency, Explainability Automated Context Enrichment AWS Cloud Enterprise

5 Meridian's Finance team counts 1,203 defaults where Risk counts 1,847. Both used the 90-day rule. What else did Finance require?

That the loan be written down in the ledger That the borrower be contacted by Collections That the loan be over a minimum amount That the loan be unsecured

6 Which statement about an ontology-driven semantic layer is true, compared with a BI semantic layer?

It stores a copy of the data for faster access It can derive facts you never explicitly stated, and detect contradictions It replaces the need for SQL anywhere in the stack It only works on AWS