Solution examples 2023

Task 2: RDF and SHACL

Questions 51-54:
RDF: Add triples TBD

For questions 55-60:

*** SHACL examples - includes answers to the exam questions

@prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> . 
@prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> . 
@prefix xsd: <http://www.w3.org/2001/XMLSchema#> . 
@prefix owl: <http://www.w3.org/2002/07/owl#> . 
@prefix dc: <http://purl.org/dc/elements/1.1/> .
@prefix foaf: <http://xmlns.com/foaf/0.1/> .
@prefix sh: <http://www.w3.org/ns/shacl#> .
@prefix : <http://info216.uib.no/movies/> .

Questions 55 and 56:

:DirectorShape a sh:NodeShape ;
    sh:targetClass :Director ;
    # A Director must have exactly one foaf:name of type xsd:string.
    sh:property [
        sh:path foaf:name ;
        sh:minCount 1 ;      # question 55
        sh:maxCount 1 ;      # question 55
        sh:type xsd:string   # question 56
    ] ;

Question 57:

    # A Director must be the director of at least one Movie.
    sh:property [
        sh:path :director_of ;
        sh:minCount 1 ;
        sh:class :Movie
    ] .

Question 58:

:ActorShape a sh:NodeShape ;
    sh:targetClass :Actor ;
    # If an actor is an actor in a resource, that resource must be a movie.
    sh:property [
        sh:path :actor_in ;
        sh:class :Movie
    ] ;

Question 59:

:ActorShape a sh:NodeShape ;
    sh:targetClass :Actor ;
    # If an actor plays a role that is a role in some resource, that resource must be a movie.
    sh:property [
        sh:path ( :plays_role :role_in ) ;
        sh:qualifiedValueShape [ sh:path :actor_in ] ;
        sh:class :Movie
    ] .

Question 60:

:MovieShape a sh:NodeShape ;
    sh:targetClass :Movie ;
    # A movie must be directed by at least one director or acted in by at least one actor.
    sh:or (
        [ sh:property [
            sh:path [ sh:inversePath :actor_in ] ;
            sh:minCount 1 ;
        ] ]
        [ sh:property [
            sh:path [ sh:inversePath :director_of ] ;
            sh:minCount 1 ;
        ] ] 
    ) .

Task 3: RDFS rules and OWL expressions

RDFS rules

Question 61: A resource that is a director_of something is a director.

:director_of rdfs:domain :Director .

Question 62: A resource that something else is a director_of is a movie.

:director_of rdfs:range :Movie .

Question 63: The year of something has type xsd:year.

:year rdfs:range xsd:year .

Question 64: An actor is a Person.

:Actor rdfs:subClassOf foaf:Person .

Question 65: A director is a person.

:Director rdfs:subClassOf foaf:Person .

OWL axioms

Question 66: Nothing can be both a person and a movie.

:Person owl:disjointWith :Movie .

Question 67: Nothing can be more than one of a person, a role, or a movie.

[] a owl:DisjointClass ;
    owl:disjointClasses ( :Person :Role :Movie ) .

Question 68: Something that plays in at least one Movie is an Actor.

:Actor rdfs:subClassOf [
    a owl:Restriction ;
    owl:onProperty :play_in ;
    owl:someValueFrom owl:Thing
]

Question 69: A LeadActor is an Actor that plays at least one LeadRole.

:LeadActor rdfs:subClassOf :Actor, [
    a owl:Restriction ;
    owl:onProperty :plays_role ;
    owl:someValueFrom :LeadRole .
] .

Task 4: SPARQL queries

Question 70: Count the number of movies that are represented in the graph.

SELECT (COUNT(?movie) AS ?count) WHERE {
    ?movie rdf:type :Movie
}

Question 71: List the titles and years of all movies.

SELECT ?title ?year WHERE {
    ?movie rdf:type :Movie ;
        dc:title ?title ;
        dc:year ?year 
}

Question 72: List the titles and years of all movies since 2000.

SELECT ?title ?year WHERE {
    ?movie rdf:type :Movie ;
        dc:title ?title ;
        dc:year ?year 
    FILTER (INTEGER(?year) >= 2000)
}

or:

SELECT ?title ?year WHERE {
    ?movie rdf:type :Movie ;
        dc:title ?title ;
        dc:year ?year 
    FILTER (?year >= "2000"^^xsd:year)
}

Question 73: List the titles and years of all movies sorted first by year, then by name.

SELECT ?title ?year WHERE {
    ?movie rdf:type :Movie ;
        dc:title ?title ;
        dc:year ?year 
}
ORDER BY ?year, ?name

Question 74: Count the number of movies for each year with more than one movie.

SELECT ?year (COUNT(?movie) AS ?count) WHERE {
    ?movie rdf:type :Movie ;
        dc:year ?year 
}
GROUP BY ?year
HAVING ?count > 1

Question 75: List the names of all persons that are both directors and actors.

SELECT ?name WHERE {
    ?person (:plays_in & :director_of) / rdf:type :Movie ;
        foaf:name ?name
}

Question 76: List the actor name and movie title for all lead roles.

SELECT ?name ?title WHERE {
    ?role rdf:type :LeadRole ;
        ^:plays_role / foaf:name ?name ;
        :role_in / dc:title ?title
}

Question 77: List all distinct pairs of actor names that have played lead roles in the same movies.

SELECT ?name1 ?name2 WHERE {
    ?movie rdf:type :Movie ;
        ^:?role_in ?role1, ?role2 .
    ?role1 rdf:type :LeadRole ;
        ^:plays_role / foaf:name ?name1 .
    ?role2 rdf:type :LeadRole ;
        ^:plays_role / foaf:name ?name2 .
    FILTER (STR(?name1) < STR(?name2))
}

Question 78-79:
SPARQL Update TBD.

Task 5: Programming

'Question 80:

from rdflib import Namespace, Graph, Literal, RDF, DC, FOAF, XSD

BASE_URI = 'http://example.org/'
MOVIE = Namespace(BASE_URI)

g = Graph()
g.bind('', MOVIE)

'Question 81:

def add_movie_triples(g, row):
    movie = row.to_dict()
    # example dict:
    # {'Movie': 'Pulp_Fiction', 'Director': 'Quentin_Tarantino', 'Year': 1994}
    movie_name = movie['Movie']
    director_name = movie['Director']
    movie_year = movie['Year']
    # update g with a set of triples that represent the movie and its director
    g.add((MOVIE[director_name], RDF.type, MOVIE.Director))
    g.add((MOVIE[director_name], FOAF.name, Literal(director_name)))
    g.add((MOVIE[director_name], MOVIE.director_of, MOVIE[movie_name]))
    g.add((MOVIE[movie_name], RDF.type, MOVIE.Movie))
    g.add((MOVIE[movie_name], DC.title, Literal(movie_name)))
    g.add((MOVIE[movie_name], MOVIE.year, Literal(movie_year, datatype=XSD.year)))

'Question 82:

from pyshacl import validate

SHACL_FILE = './movie-shacl.ttl'
# contains the rule for "A Movie must have exactly one dc:title."


# ...


sg = Graph()
sg.parse(SHACL_FILE, format='ttl')
r = validate(g,
        shacl_graph=sg,
        # ont_graph=og,
        inference='rdfs'
    )
val, rg, rep = r
print(rep)

'Question 83:
TBD.

'Question 84:

from owlrl import DeductiveClosure, OWLRL_Semantics

ONTOLOGY_FILE = './movie-ontology.ttl'

# ...

g.parse(ONTOLOGY_FILE)
DeductiveClosure(OWLRL_Semantics).expand(g)

print(g.serialize(format='ttl'))

'Full program

For reference, not expected exam answer:

from owlrl import DeductiveClosure, OWLRL_Semantics
import pandas as pd
from pyshacl import validate
from rdflib import Namespace, Graph, Literal, RDF, DC, FOAF, XSD


ONTOLOGY_FILE = './movie-ontology.ttl'
SHACL_FILE = './movie-shacl.ttl'
DIRECTOR_FILE = './movie-director-year.csv'
LEAD_ROLE_FILE = './movie-actor-lead-role.csv'
OTHER_ROLE_FILE = './movie-actor-other-role.csv'

BASE_URI = 'http://example.org/'
MOVIE = Namespace(BASE_URI)


def add_movie_triples(g, row):
    movie = row.to_dict()
    # example dict:
    # {'Movie': 'Pulp_Fiction', 'Director': 'Quentin_Tarantino', 'Year': 1994}
    movie_name = movie['Movie']
    director_name = movie['Director']
    movie_year = movie['Year']
    # update g with a set of triples that represent the movie and its director
    g.add((MOVIE[director_name], RDF.type, MOVIE.Director))
    g.add((MOVIE[director_name], FOAF.name, Literal(director_name)))
    g.add((MOVIE[director_name], MOVIE.director_of, MOVIE[movie_name]))
    g.add((MOVIE[movie_name], RDF.type, MOVIE.Movie))
    g.add((MOVIE[movie_name], DC.title, Literal(movie_name)))
    g.add((MOVIE[movie_name], MOVIE.year, Literal(movie_year, datatype=XSD.year)))


def add_lead_role_triples(g, row):
    movie = row.to_dict()
    # example dict:
    # {'Movie': 'Pulp_Fiction', 'Director': 'Quentin_Tarantino', 'Year': 1994}
    movie_name = movie['Movie']
    actor_name = movie['Actor']
    role_name = movie_name+'-role-'+movie['LeadRole']
    # update g with a set of triples that represent the movie and its director
    g.add((MOVIE[actor_name], RDF.type, MOVIE.Actor))
    g.add((MOVIE[actor_name], FOAF.name, Literal(actor_name)))
    g.add((MOVIE[actor_name], MOVIE.actor_in, MOVIE[movie_name]))
    g.add((MOVIE[actor_name], MOVIE.plays_role, MOVIE[role_name]))
    g.add((MOVIE[role_name], RDF.type, MOVIE.LeadRole))
    g.add((MOVIE[role_name], FOAF.name, Literal(movie['LeadRole'])))
    g.add((MOVIE[role_name], MOVIE.role_in, MOVIE[movie_name]))
    g.add((MOVIE[movie_name], RDF.type, MOVIE.Movie))


def add_other_role_triples(g, row):
    movie = row.to_dict()
    # example dict:
    # {'Movie': 'Pulp_Fiction', 'Director': 'Quentin_Tarantino', 'Year': 1994}
    movie_name = movie['Movie']
    actor_name = movie['Actor']
    role_name = movie_name+'-role-'+movie['Role']
    # update g with a set of triples that represent the movie and its director
    g.add((MOVIE[actor_name], RDF.type, MOVIE.Actor))
    g.add((MOVIE[actor_name], FOAF.name, Literal(actor_name)))
    g.add((MOVIE[actor_name], MOVIE.actor_in, MOVIE[movie_name]))
    g.add((MOVIE[actor_name], MOVIE.plays_role, MOVIE[role_name]))
    g.add((MOVIE[role_name], RDF.type, MOVIE.Role))
    g.add((MOVIE[role_name], FOAF.name, Literal(movie['Role'])))
    g.add((MOVIE[role_name], MOVIE.role_in, MOVIE[movie_name]))
    g.add((MOVIE[movie_name], RDF.type, MOVIE.Movie))


def load_movie_triples(g, fn):
    df = pd.read_csv(fn)
    df.apply(lambda row: add_movie_triples(g, row), axis=1)


def load_lead_role_triples(g, fn):
    df = pd.read_csv(fn)
    df.apply(lambda row: add_lead_role_triples(g, row), axis=1)


def load_other_role_triples(g, fn):
    df = pd.read_csv(fn)
    df.apply(lambda row: add_other_role_triples(g, row), axis=1)


g = Graph()
g.bind('', MOVIE)
load_movie_triples(g, DIRECTOR_FILE)
load_lead_role_triples(g, LEAD_ROLE_FILE)
load_other_role_triples(g, OTHER_ROLE_FILE)
print(g.serialize(format='ttl'))


sg = Graph()
sg.parse(SHACL_FILE, format='ttl')
r = validate(g,
        shacl_graph=sg,
        # ont_graph=og,
        inference='rdfs'
    )
val, rg, rep = r
print(rep)


g.parse(ONTOLOGY_FILE)
DeductiveClosure(OWLRL_Semantics).expand(g)

print(g.serialize(format='ttl'))