Skip to content

Commit 75899bc

Browse files
dgarijodgarijo
authored
Merge pull request #30 from pggil2019-design/master
Assignment 4 Pablo
2 parents 27ed10e + 809098f commit 75899bc

5 files changed

Lines changed: 639 additions & 0 deletions

File tree

‎Assignment4/Assignment4Pablo_Gil_24C027/task06.py‎

Lines changed: 138 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,138 @@
1+
# -*- coding: utf-8 -*-
2+
"""Task06.ipynb
3+
4+
Automatically generated by Colab.
5+
6+
Original file is located at
7+
https://colab.research.google.com/drive/1Rf1_3J-aHPNigtrFH4R4_EcugyF66LMI
8+
9+
**Task 06: Modifying RDF(s)**
10+
"""
11+
12+
#!pip install rdflib
13+
import urllib.request
14+
url = 'https://raw.githubusercontent.com/FacultadInformatica-LinkedData/Curso2025-2026/refs/heads/master/Assignment4/course_materials/python/validation.py'
15+
urllib.request.urlretrieve(url, 'validation.py')
16+
github_storage = "https://raw.githubusercontent.com/FacultadInformatica-LinkedData/Curso2025-2026/master/Assignment4/course_materials"
17+
18+
"""Import RDFLib main methods"""
19+
20+
from rdflib import Graph, Namespace, Literal, XSD
21+
from rdflib.namespace import RDF, RDFS
22+
from validation import Report
23+
g = Graph()
24+
g.namespace_manager.bind('ns', Namespace("http://somewhere#"), override=False)
25+
r = Report()
26+
27+
"""Create a new class named Researcher"""
28+
29+
ns = Namespace("http://mydomain.org#")
30+
g.add((ns.Researcher, RDF.type, RDFS.Class))
31+
for s, p, o in g:
32+
print(s,p,o)
33+
34+
"""**Task 6.0: Create new prefixes for "ontology" and "person" as shown in slide 14 of the Slidedeck 01a.RDF(s)-SPARQL shown in class.**"""
35+
36+
# this task is validated in the next step
37+
ontology = Namespace("http://www.oeg-upm.net/Ontology#")
38+
person = Namespace("http://oeg.fi.upm.es/def/people#")
39+
40+
g.namespace_manager.bind('ontology', ontology)
41+
g.namespace_manager.bind('person', person)
42+
43+
"""**TASK 6.1: Reproduce the taxonomy of classes shown in slide 34 in class (all the classes under "Vocabulario", Slidedeck: 01a.RDF(s)-SPARQL). Add labels for each of them as they are in the diagram (exactly) with no language tags. Remember adding the correct datatype (xsd:String) when appropriate**
44+
45+
"""
46+
47+
# TO DO
48+
g.add((person.Person, RDF.type, RDFS.Class))
49+
g.add((person.Person, RDFS.label, Literal("Person", datatype=XSD.string)))
50+
g.add((person.Professor, RDF.type, RDFS.Class))
51+
g.add((person.Professor, RDFS.subClassOf, person.Person))
52+
g.add((person.Professor, RDFS.label, Literal("Professor", datatype=XSD.string)))
53+
g.add((person.FullProfessor, RDF.type, RDFS.Class))
54+
g.add((person.FullProfessor, RDFS.subClassOf, person.Professor))
55+
g.add((person.FullProfessor, RDFS.label, Literal("FullProfessor", datatype=XSD.string)))
56+
g.add((person.AssociateProfessor, RDF.type, RDFS.Class))
57+
g.add((person.AssociateProfessor, RDFS.subClassOf, person.Professor))
58+
g.add((person.AssociateProfessor, RDFS.label, Literal("AssociateProfessor", datatype=XSD.string)))
59+
g.add((person.InterimAssociateProfessor, RDF.type, RDFS.Class))
60+
g.add((person.InterimAssociateProfessor, RDFS.subClassOf, person.AssociateProfessor))
61+
g.add((person.InterimAssociateProfessor, RDFS.label, Literal("InterimAssociateProfessor", datatype=XSD.string)))
62+
63+
# Visualize the results
64+
for s, p, o in g:
65+
print(s,p,o)
66+
67+
# Validation. Do not remove
68+
r.validate_task_06_01(g)
69+
70+
"""**TASK 6.2: Add the 3 properties shown in slide 36. Add labels for each of them (exactly as they are in the slide, with no language tags), and their corresponding domains and ranges using RDFS. Remember adding the correct datatype (xsd:String) when appropriate. If a property has no range, make it a literal (string)**"""
71+
72+
# TO DO
73+
#hasName
74+
g.add((person.hasName, RDF.type, RDF.Property))
75+
g.add((person.hasName, RDFS.domain, person.Person))
76+
g.add((person.hasName, RDFS.range, RDFS.Literal))
77+
g.add((person.hasName, RDFS.label, Literal("hasName", datatype=XSD.string)))
78+
#hasColleague
79+
g.add((person.hasColleague, RDF.type, RDF.Property))
80+
g.add((person.hasColleague, RDFS.domain, person.Person))
81+
g.add((person.hasColleague, RDFS.range, person.Person))
82+
g.add((person.hasColleague, RDFS.label, Literal("hasColleague", datatype=XSD.string)))
83+
#hasHomePage
84+
g.add((person.hasHomePage, RDF.type, RDF.Property))
85+
g.add((person.hasHomePage, RDFS.domain, person.FullProfessor))
86+
g.add((person.hasHomePage, RDFS.range, RDFS.Literal))
87+
g.add((person.hasHomePage, RDFS.label, Literal("hasHomePage", datatype=XSD.string)))
88+
# Visualize the results
89+
for s, p, o in g:
90+
print(s,p,o)
91+
92+
# Validation. Do not remove
93+
r.validate_task_06_02(g)
94+
95+
"""**TASK 6.3: Create the individuals shown in slide 36 under "Datos". Link them with the same relationships shown in the diagram."**"""
96+
97+
# TO DO
98+
data = Namespace("http://oeg.fi.upm.es/resource/person/")
99+
g.namespace_manager.bind('data', data)
100+
101+
g.add((data.Oscar, RDF.type, person.AssociateProfessor))
102+
g.add((data.Oscar, person.hasColleague, data.Asun))
103+
g.add((data.Oscar, person.hasName, Literal("Oscar Corcho García", datatype=XSD.string)))
104+
g.add((data.Oscar, RDFS.label, Literal("Oscar", datatype=XSD.string)))
105+
g.add((data.Asun, RDF.type, person.FullProfessor))
106+
g.add((data.Asun, person.hasColleague, data.Raul))
107+
g.add((data.Asun, person.hasHomePage, Literal("http://www.oeg-upm.net/", datatype=XSD.string)))
108+
g.add((data.Asun, RDFS.label, Literal("Asun", datatype=XSD.string)))
109+
g.add((data.Raul, RDF.type, person.InterimAssociateProfessor))
110+
g.add((data.Raul, RDFS.label, Literal("Raul", datatype=XSD.string)))
111+
112+
# Visualize the results
113+
for s, p, o in g:
114+
print(s,p,o)
115+
116+
r.validate_task_06_03(g)
117+
118+
"""**TASK 6.4: Add to the individual person:Oscar the email address, given and family names. Use the properties already included in example 4 to describe Jane and John (https://raw.githubusercontent.com/FacultadInformatica-LinkedData/Curso2025-2026/master/Assignment4/course_materials/rdf/example4.rdf). Do not import the namespaces, add them manually**
119+
120+
"""
121+
122+
# TO DO
123+
foaf = Namespace("http://xmlns.com/foaf/0.1/")
124+
vcard = Namespace("http://www.w3.org/2001/vcard-rdf/3.0/")
125+
126+
g.namespace_manager.bind('foaf', foaf)
127+
g.namespace_manager.bind('vcard', vcard)
128+
129+
g.add((data.Oscar, vcard.Given, Literal("Oscar", datatype=XSD.string)))
130+
g.add((data.Oscar, vcard.Family, Literal("Corcho García", datatype=XSD.string)))
131+
g.add((data.Oscar, foaf.email, Literal("ocorcho@fi.upm.es", datatype=XSD.string)))
132+
# Visualize the results
133+
for s, p, o in g:
134+
print(s,p,o)
135+
136+
# Validation. Do not remove
137+
r.validate_task_06_04(g)
138+
r.save_report("_Task_06")

‎Assignment4/Assignment4Pablo_Gil_24C027/task07.py‎

Lines changed: 152 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,152 @@
1+
# -*- coding: utf-8 -*-
2+
"""Task07.ipynb
3+
4+
Automatically generated by Colab.
5+
6+
Original file is located at
7+
https://colab.research.google.com/drive/1WdK0g7IDVNpTbV3nfDcyhYiGfNwj5JPL
8+
9+
**Task 07: Querying RDF(s)**
10+
"""
11+
12+
#!pip install rdflib
13+
import urllib.request
14+
url = 'https://raw.githubusercontent.com/FacultadInformatica-LinkedData/Curso2025-2026/refs/heads/master/Assignment4/course_materials/python/validation.py'
15+
urllib.request.urlretrieve(url, 'validation.py')
16+
github_storage = "https://raw.githubusercontent.com/FacultadInformatica-LinkedData/Curso2025-2026/master/Assignment4/course_materials"
17+
18+
from validation import Report
19+
20+
"""First let's read the RDF file"""
21+
22+
from rdflib import Graph, Namespace, Literal
23+
from rdflib.namespace import RDF, RDFS
24+
# Do not change the name of the variables
25+
g = Graph()
26+
g.namespace_manager.bind('ns', Namespace("http://somewhere#"), override=False)
27+
g.parse(github_storage+"/rdf/data06.ttl", format="TTL")
28+
report = Report()
29+
30+
"""**TASK 7.1a: For all classes, list each classURI. If the class belogs to another class, then list its superclass.**
31+
**Do the exercise in RDFLib returning a list of Tuples: (class, superclass) called "result". If a class does not have a super class, then return None as the superclass**
32+
"""
33+
34+
# TO DO
35+
# Visualize the results
36+
result = [] #list of tuples
37+
for subject in g.subjects(RDF.type, RDFS.Class):
38+
superclass=None
39+
for super_class in g.objects(subject, RDFS.subClassOf):
40+
superclass=super_class
41+
result.append((subject, superclass))
42+
for r in result:
43+
print(r)
44+
#Para mostrar los nombres reales
45+
for c, sc in result:
46+
short_c = g.namespace_manager.normalizeUri(c)
47+
short_sc = g.namespace_manager.normalizeUri(sc) if sc else None
48+
print(short_c, short_sc)
49+
50+
## Validation: Do not remove
51+
report.validate_07_1a(result)
52+
53+
"""**TASK 7.1b: Repeat the same exercise in SPARQL, returning the variables ?c (class) and ?sc (superclass)**"""
54+
55+
query = """
56+
SELECT ?c ?sc WHERE {
57+
?c a rdfs:Class.
58+
OPTIONAL { ?c rdfs:subClassOf ?sc.}
59+
}
60+
"""
61+
62+
for r in g.query(query):
63+
print(r.c, r.sc)
64+
65+
## Validation: Do not remove
66+
report.validate_07_1b(query,g)
67+
68+
"""**TASK 7.2a: List all individuals of "Person" with RDFLib (remember the subClasses). Return the individual URIs in a list called "individuals"**
69+
70+
"""
71+
72+
ns = Namespace("http://oeg.fi.upm.es/def/people#")
73+
person_classes=[ns.Person, ns.Professor, ns.Student, ns.FullProfessor,
74+
ns.AssociateProfessor, ns.InterimAssociateProfessor]
75+
# variable to return
76+
individuals = []
77+
for c in person_classes:
78+
for subject in g.subjects(RDF.type, c):
79+
individuals.append(subject)
80+
# visualize results
81+
for i in individuals:
82+
print(i)
83+
84+
# validation. Do not remove
85+
86+
report.validate_07_02a(individuals)
87+
88+
"""**TASK 7.2b: Repeat the same exercise in SPARQL, returning the individual URIs in a variable ?ind**"""
89+
90+
query = """
91+
PREFIX ns: <http://oeg.fi.upm.es/def/people#>
92+
93+
SELECT ?ind WHERE {
94+
?ind rdf:type/rdfs:subClassOf* ns:Person .
95+
}
96+
"""
97+
98+
for r in g.query(query):
99+
print(r.ind)
100+
# Visualize the results
101+
102+
## Validation: Do not remove
103+
report.validate_07_02b(g, query)
104+
105+
"""**TASK 7.3: List the name and type of those who know Rocky (in SPARQL only). Use name and type as variables in the query**"""
106+
107+
query = """
108+
PREFIX ns: <http://oeg.fi.upm.es/def/people#>
109+
SELECT ?name ?type WHERE {
110+
?x ns:knows ns:Rocky .
111+
?x rdfs:label ?name .
112+
?x a ?type .
113+
}
114+
"""
115+
# TO DO
116+
# Visualize the results
117+
for r in g.query(query):
118+
print(r.name, r.type)
119+
120+
## Validation: Do not remove
121+
report.validate_07_03(g, query)
122+
123+
"""**Task 7.4: List the name of those entities who have a colleague with a dog, or that have a collegue who has a colleague who has a dog (in SPARQL). Return the results in a variable called name**"""
124+
125+
query = """
126+
PREFIX ns: <http://oeg.fi.upm.es/def/people#>
127+
SELECT DISTINCT ?name WHERE {
128+
{
129+
?x ns:hasColleague ?y .
130+
?y ns:ownsPet ?pet .
131+
?pet a ns:Animal .
132+
}
133+
UNION
134+
{
135+
?x ns:hasColleague ?y .
136+
?y ns:hasColleague ?z .
137+
?z ns:ownsPet ?pet .
138+
?pet a ns:Animal .
139+
}
140+
?x rdfs:label ?name .
141+
}
142+
"""
143+
144+
for r in g.query(query):
145+
print(r.name)
146+
147+
# TO DO
148+
# Visualize the results
149+
150+
## Validation: Do not remove
151+
report.validate_07_04(g,query)
152+
report.save_report("_Task_07")

‎Assignment4/Assignment4Pablo_Gil_24C027/task08.py‎

Lines changed: 51 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,51 @@
1+
# -*- coding: utf-8 -*-
2+
"""Task08.ipynb
3+
4+
Automatically generated by Colab.
5+
6+
Original file is located at
7+
https://colab.research.google.com/drive/13psLr0xZv7Y04g2XaBkB3w7iity008RI
8+
9+
**Task 08: Completing missing data**
10+
"""
11+
12+
#!pip install rdflib
13+
github_storage = "https://raw.githubusercontent.com/FacultadInformatica-LinkedData/Curso2021-2022/master/Assignment4/course_materials"
14+
15+
from rdflib import Graph, Namespace, Literal, URIRef
16+
g1 = Graph()
17+
g2 = Graph()
18+
g1.parse(github_storage+"/rdf/data01.rdf", format="xml")
19+
g2.parse(github_storage+"/rdf/data02.rdf", format="xml")
20+
21+
"""Tarea: lista todos los elementos de la clase Person en el primer grafo (data01.rdf) y completa los campos (given name, family name y email) que puedan faltar con los datos del segundo grafo (data02.rdf). Puedes usar consultas SPARQL o iterar el grafo, o ambas cosas."""
22+
23+
from rdflib.namespace import RDF, RDFS
24+
25+
vcard = Namespace("http://www.w3.org/2001/vcard-rdf/3.0#")
26+
data = Namespace("http://data.org#")
27+
28+
g1.namespace_manager.bind('vcard', vcard)
29+
g1.namespace_manager.bind('data', data)
30+
g2.namespace_manager.bind('vcard', vcard)
31+
g2.namespace_manager.bind('data', data)
32+
33+
for s in g1.subjects(RDF.type, data.Person):
34+
print (s)
35+
36+
for person in g1.subjects(RDF.type, data.Person):
37+
if not g1.value(person, vcard.Given):
38+
nombre = g2.value(person, vcard.Given)
39+
if nombre:
40+
g1.add((person, vcard.Given, nombre))
41+
if not g1.value(person, vcard.Family):
42+
apellido = g2.value(person, vcard.Family)
43+
if apellido:
44+
g1.add((person, vcard.Family, apellido))
45+
if not g1.value(person, vcard.EMAIL):
46+
email = g2.value(person, vcard.EMAIL)
47+
if email:
48+
g1.add((person, vcard.EMAIL, email))
49+
50+
for s, p, o in g1:
51+
print(s,p,o)

‎Assignment4/Assignment4Pablo_Gil_24C027/task09.py‎

Lines changed: 40 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,40 @@
1+
# -*- coding: utf-8 -*-
2+
"""Task09.ipynb
3+
4+
Automatically generated by Colab.
5+
6+
Original file is located at
7+
https://colab.research.google.com/drive/17TeGusVRAOJZAhmEbSjv5iHoDGtfiz2Q
8+
9+
**Task 09: Data linking**
10+
"""
11+
12+
#!pip install rdflib
13+
github_storage = "https://raw.githubusercontent.com/FacultadInformatica-LinkedData/Curso2021-2022/master/Assignment4/course_materials/"
14+
15+
from rdflib import Graph, Namespace, Literal, URIRef
16+
g1 = Graph()
17+
g2 = Graph()
18+
g3 = Graph()
19+
g1.parse(github_storage+"rdf/data03.rdf", format="xml")
20+
g2.parse(github_storage+"rdf/data04.rdf", format="xml")
21+
22+
"""Busca individuos en los dos grafos y enlázalos mediante la propiedad OWL:sameAs, inserta estas coincidencias en g3. Consideramos dos individuos iguales si tienen el mismo apodo y nombre de familia. Ten en cuenta que las URI no tienen por qué ser iguales para un mismo individuo en los dos grafos."""
23+
24+
from rdflib.namespace import RDF, RDFS
25+
owl = Namespace("http://www.w3.org/2002/07/owl#")
26+
vcard = Namespace("http://www.w3.org/2001/vcard-rdf/3.0#")
27+
28+
for s1 in g1.subjects(RDF.type, None):
29+
given1 = next(g1.objects(s1, vcard.Given), None)
30+
family1=next(g1.objects(s1, vcard.Family), None)
31+
if given1 and family1:
32+
for s2 in g2.subjects(RDF.type, None):
33+
given2 = next(g2.objects(s2, vcard.Given), None)
34+
family2 = next(g2.objects(s2, vcard.Family), None)
35+
if given2 and family2:
36+
if str(given1) == str(given2) and str(family1) == str(family2):
37+
g3.add((s1, owl.sameAs, s2))
38+
39+
for s, p, o in g3:
40+
print(s, p, s)

0 commit comments

Comments
 (0)