4. Asking questions with SPARQL | Semantic Layer course

Section 1

Matching a shape, not joining tables

Next section

Section 2

Watching patterns bind

Next section

Section 3

OPTIONAL, UNION, and the cost of getting them backwards

Next section

Section 4

Aggregates and the four query forms

Next section

Section 5

Property paths, and the query SQL hates

Next section

Section 6

Closing the Tuesday loop

Hands on

Answer the CRO's question three ways

30 min

Run all three default definitions as SPARQL against Fuseki and reproduce the counts 1,847, 4,102 and 1,203 from one unchanged graph.

make lab-up            # once, starts Postgres + Fuseki + Ontop
make lab lab04-sparql-drills

Full instructions, starter files and the solution are in labs/modules/lab04-sparql-drills/README.md. The lab checks its own work, so you will know if it passed.

Check yourself

Quiz

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

1 How does a SPARQL WHERE block differ from a SQL FROM and JOIN clause?

SPARQL requires you to specify the join order for performance SPARQL describes the shape of the answer as triple patterns with variables, and the engine finds every subgraph that fits SPARQL can only query one predicate at a time There is no difference beyond syntax

2 You query all loans and their collateral without OPTIONAL. What happens to Daniel's unsecured loan L-2087?

It appears with collateral bound to zero It disappears from the results entirely, with no error or warning The query fails with a missing property error It appears with collateral bound to an empty string

3 Which two query forms would you use for these jobs: deciding whether one specific loan meets a rule, and writing inference results back into the graph? (pick two)

Select all that apply.

ASK, for the single true or false decision DESCRIBE, for the inference results CONSTRUCT, for producing new triples to load SELECT, for both

4 In SPARQL, what does mrb:relatedParty+ mean?

Match the predicate relatedParty and any predicate whose name starts with it Follow relatedParty one or more times, transitively, with cycles handled by the engine Require at least one relatedParty triple but do not follow further Concatenate the results of two relatedParty queries

5 Meridian's Finance definition of default counts 1,203 loans where Risk counts 1,847. In the Finance SPARQL query, what produces that difference?

A FILTER checking that the write-off date IS NOT NULL An extra required triple pattern for mrb:writeOffDate, which excludes loans that have no such triple A HAVING clause on the aggregate A UNION with the written-off loans graph

6 You write COUNT(?collateral) where ?collateral is bound inside an OPTIONAL block, intending to count loans per branch. What do you get?

The number of loans per branch, since OPTIONAL preserves every row The number of loans that happen to have collateral, presented as if it were a loan count Zero for every branch An error, because aggregates cannot reference OPTIONAL variables

7 What is the SERVICE keyword for?

Calling a REST API from inside a query Sending part of a query to another SPARQL endpoint and merging the results into your bindings Registering the query as a scheduled job Declaring which named graph to query