5. Building Meridian's ontology | Semantic Layer course

Section 1

Two kinds of triple in one graph

Next section

Section 2

Start from the questions, not the classes

Next section

Section 3

RDFS: the cheap half of the vocabulary

Next section

Section 4

OWL: the axioms that let a machine draw conclusions

Next section

Section 5

The finished hierarchy

Next section

Section 6

Governing a file that decides who gets a loan

Next section

Section 7

What the T-Box is now capable of

Hands on

Write the Meridian T-Box

35 min

Author meridian.ttl with the class hierarchy, properties with domain and range, disjointness and functional axioms; load it into Fuseki; run a SPARQL query that lists every class and its parent.

make lab-up            # once, starts Postgres + Fuseki + Ontop
make lab lab05-build-ontology

Full instructions, starter files and the solution are in labs/modules/lab05-build-ontology/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 Which triple belongs to the T-Box rather than the A-Box?

loan:L-1042 mrb:outstandingBalance 291450.00 mrb:MortgageLoan rdfs:subClassOf mrb:SecuredLoan cust:C-1003 mrb:relatedParty cust:C-1001 loan:L-4590 mrb:daysPastDue 104

2 Why is competency-question-first modelling recommended over starting with a class diagram?

Competency questions run faster in SPARQL Every class you build traces to a question somebody asked, and the questions become your regression tests OWL requires a list of questions before you can declare a class It is the only way to align with FIBO

3 Which two statements about skos:altLabel are correct? (pick two)

Select all that apply.

It lets "NPA", "gone bad" and "delinquent loan" all resolve to mrb:NonPerformingLoan It changes the logical meaning of the class for the reasoner It helps an agent map a user's actual phrasing onto the right concept It is required on every OWL class or the ontology is invalid

4 Meridian declares mrb:borrower as an owl:FunctionalProperty. What does the reasoner conclude if loan L-2087 is asserted to have both Daniel Okafor and Priya Raman as borrower?

The second assertion is rejected and never stored It concludes Daniel Okafor and Priya Raman must be the same individual, and reports an inconsistency once they are declared different Nothing, functional properties only apply to datatype properties It silently keeps the most recently written value

5 What is the practical difference between using rdfs:subClassOf and owl:equivalentClass with a restriction to describe mrb:DefaultedLoan?

subClassOf is faster to reason over, equivalentClass is slower but identical in meaning subClassOf states a necessary condition only; equivalentClass makes the restriction a full definition, so matching individuals get classified automatically equivalentClass only works in OWL 2 DL, subClassOf works everywhere There is no difference, the two are synonyms in Turtle

6 Which is the honest position on adopting FIBO for Meridian?

Import all of FIBO and model nothing yourself Ignore FIBO, since a mid-size retail bank has no external reporting obligations Align a fragment: map the boundary terms like Loan and Party to FIBO, and keep Meridian's own policy concepts in the mrb namespace Wait until FIBO publishes a retail banking edition

7 Meridian splits mrb:DefaultedLoan into three separately named default classes. Under semantic versioning, what version bump is that?

Patch, because no data changed Minor, because classes were added Major, because an existing class changed meaning or disappeared and existing queries break No bump; class changes are tracked by owl:versionInfo only