Lab Solutions: Difference between revisions

From info216
No edit summary
(Proposed solution for Lab 7 RDFS)
(38 intermediate revisions by 4 users not shown)
Line 1: Line 1:
This page will be updated with Python examples related to the lectures and labs. We will add more examples after each lab has ended. The first examples will use Python's RDFlib. We will introduce other relevant libraries later.
This page will be updated with Python examples related to the labs as the course progresses.


==Getting started==
=Examples from the lectures=
 
==Lecture 1: Introduction to KGs==
Turtle example:
<syntaxhighlight>
<syntaxhighlight>
@prefix ex: <http://example.org/> .
ex:Roger_Stone
    ex:name "Roger Stone" ;
    ex:occupation ex:lobbyist ;
    ex:significant_person ex:Donald_Trump .
ex:Donald_Trump
    ex:name "Donald Trump" .
</syntaxhighlight>
==Lecture 2: RDF==
Blank nodes for anonymity, or when we have not decided on a URI:
<syntaxhighlight lang="Python">
from rdflib import Graph, Namespace, Literal, BNode, RDF, RDFS, DC, FOAF, XSD


from rdflib.collection import Collection
EX = Namespace('http://example.org/')
from rdflib import Graph, Namespace, Literal, URIRef
from rdflib.namespace import RDF, FOAF, XSD


g = Graph()
g = Graph()
EX = Namespace('http://EXample.org/')
g.bind('ex', EX) # this is why the line '@prefix ex: <http://example.org/> .'
RL = Namespace('http://purl.org/vocab/relationship/')
                  # and the 'ex.' prefix are used when we print out Turtle later
DBO = Namespace('https://dbpedia.org/ontology/')
DBR = Namespace('https://dbpedia.org/page/')


g.namespace_manager.bind('exampleURI', EX)
robertMueller = BNode()
g.namespace_manager.bind('relationship', RL)
g.add((robertMueller, RDF.type, EX.Human))
g.namespace_manager.bind('dbpediaOntology', DBO)
g.add((robertMueller, FOAF.name, Literal('Robert Mueller', lang='en')))
g.namespace_manager.bind('dbpediaPage', DBR)
g.add((robertMueller, EX.position_held, Literal('Director of the Federal Bureau of Investigation', lang='en')))


g.add((EX.Cade, RDF.type, FOAF.Person))
print(g.serialize(format='turtle'))
g.add((EX.Mary, RDF.type, FOAF.Person))
</syntaxhighlight>
g.add((EX.Cade, RL.spouseOf, EX.Mary)) # a symmetrical relation from an established namespace
g.add((DBR.France, DBO.capital, DBR.Paris))
g.add((EX.Cade, FOAF.age, Literal(27)))
g.add((EX.Mary, FOAF.age, Literal('26', datatype=XSD.int)))
Collection (g, EX.MaryInterests, [EX.hiking, EX.choclate, EX.biology])
g.add((EX.Mary, EX.hasIntrest, EX.MaryInterests))
g.add((EX.Mary, RDF.type, EX.student))
g.add((DBO.capital, EX.range, EX.city))
g.add((EX.Mary, RDF.type, EX.kind))
g.add((EX.Cade, RDF.type, EX.kindPerson))


#hobbies = ['hiking', 'choclate', 'biology']
Blank nodes used to group related properties:
#for i in hobbies:
<syntaxhighlight>
#    g.add((EX.Mary, FOAF.interest, EX[i]))
from rdflib import Graph, Namespace, Literal, BNode, RDF, RDFS, DC, FOAF, XSD


print(g.serialize(format="turtle"))
EX = Namespace('http://example.org/')
</syntaxhighlight>
==RDFlib==


g = Graph()
g.bind('ex', EX)


==SPARQL - Blazegraph==
# This is a task in Exercise 2


print(g.serialize(format='turtle'))
</syntaxhighlight>


==SPARQL - RDFlib==
Literals:
<syntaxhighlight>
from rdflib import Graph, Namespace, Literal, BNode, RDF, RDFS, DC, FOAF, XSD


import requests
EX = Namespace('http://example.org/')
from rdflib import FOAF, Namespace, Literal, RDF, Graph, TURTLE


r = requests.get('http://api.open-notify.org/astros.json').json()
g = Graph()
g = Graph()
g.bind('ex', EX)
EX = Namespace('http://EXample.org/')
g.bind("ex", EX)


for item in r['people']:
g.add((EX.Robert_Mueller, RDF.type, EX.Human))
    craft = item['craft'].replace(" ","_")
g.add((EX.Robert_Mueller, FOAF.name, Literal('Robert Mueller', lang='en')))
    person = item['name'].replace(" ","_")
g.add((EX.Robert_Mueller, FOAF.name, Literal('رابرت مولر', lang='fa')))
    g.add((EX[person], EX.onCraft, EX[craft]))
g.add((EX.Robert_Mueller, DC.description, Literal('sixth director of the FBI', datatype=XSD.string)))
    g.add((EX[person], RDF.type, FOAF.Person))
g.add((EX.Robert_Mueller, EX.start_time, Literal(2001, datatype=XSD.integer)))
    g.add((EX[person], FOAF.name, Literal(item['name'])))
    g.add((EX[craft], FOAF.name, Literal(item['craft'])))
res = g.query("""
    CONSTRUCT {?person1 foaf:knows ?person2}
    WHERE {
        ?person1 ex:onCraft ?craft .
        ?person2 ex:onCraft ?craft .
        }
""")


for triplet in res:
print(g.serialize(format='turtle'))
    if (triplet[0] != triplet[2]):
</syntaxhighlight>
        g.add((triplet))
       
print(g.serialize(format="turtle"))
==Web APIs and JSON-LD==


Alternative container (open):
<syntaxhighlight>
from rdflib import Graph, Namespace, Literal, BNode, RDF, RDFS, DC, FOAF, XSD


==Semantic lifting - CSV==
EX = Namespace('http://example.org/')


g = Graph()
g.bind('ex', EX)


==RDFS==
muellerReportArchives = BNode()
g.add((muellerReportArchives, RDF.type, RDF.Alt))


archive1 = 'https://archive.org/details/MuellerReportVolume1Searchable/' \
                    'Mueller%20Report%20Volume%201%20Searchable/'
archive2 = 'https://edition.cnn.com/2019/04/18/politics/full-mueller-report-pdf/index.html'
archive3 = 'https://www.politico.com/story/2019/04/18/mueller-report-pdf-download-text-file-1280891'


g.add((muellerReportArchives, RDFS.member, Literal(archive1, datatype=XSD.anyURI)))
g.add((muellerReportArchives, RDFS.member, Literal(archive2, datatype=XSD.anyURI)))
g.add((muellerReportArchives, RDFS.member, Literal(archive3, datatype=XSD.anyURI)))


==Getting started==
g.add((EX.Mueller_Report, RDF.type, FOAF.Document))
g.add((EX.Mueller_Report, DC.contributor, EX.Robert_Mueller))
g.add((EX.Mueller_Report, SCHEMA.archivedAt, muellerReportArchives))


print(g.serialize(format='turtle'))
</syntaxhighlight>


===Printing the triples of the Graph in a readable way===
Sequence container (open):
<syntaxhighlight>
<syntaxhighlight>
# The turtle format has the purpose of being more readable for humans.
from rdflib import Graph, Namespace, Literal, BNode, RDF, RDFS, DC, FOAF, XSD
print(g.serialize(format="turtle"))
</syntaxhighlight>


===Coding Tasks Lab 1===
EX = Namespace('http://example.org/')
<syntaxhighlight>
from rdflib import Graph, Namespace, URIRef, BNode, Literal
from rdflib.namespace import RDF, FOAF, XSD


g = Graph()
g = Graph()
ex = Namespace("http://example.org/")
g.bind('ex', EX)
 
donaldTrumpSpouses = BNode()
g.add((donaldTrumpSpouses, RDF.type, RDF.Seq))
g.add((donaldTrumpSpouses, RDF._1, EX.IvanaTrump))
g.add((donaldTrumpSpouses, RDF._2, EX.MarlaMaples))
g.add((donaldTrumpSpouses, RDF._3, EX.MelaniaTrump))
 
g.add((EX.Donald_Trump, SCHEMA.spouse, donaldTrumpSpouses))
 
print(g.serialize(format='turtle'))
</syntaxhighlight>


g.add((ex.Cade, ex.married, ex.Mary))
Collection (closed list):
g.add((ex.France, ex.capital, ex.Paris))
<syntaxhighlight>
g.add((ex.Cade, ex.age, Literal("27", datatype=XSD.integer)))
from rdflib import Graph, Namespace, Literal, BNode, RDF, RDFS, DC, FOAF, XSD
g.add((ex.Mary, ex.age, Literal("26", datatype=XSD.integer)))
g.add((ex.Mary, ex.interest, ex.Hiking))
g.add((ex.Mary, ex.interest, ex.Chocolate))
g.add((ex.Mary, ex.interest, ex.Biology))
g.add((ex.Mary, RDF.type, ex.Student))
g.add((ex.Paris, RDF.type, ex.City))
g.add((ex.Paris, ex.locatedIn, ex.France))
g.add((ex.Cade, ex.characteristic, ex.Kind))
g.add((ex.Mary, ex.characteristic, ex.Kind))
g.add((ex.Mary, RDF.type, FOAF.Person))
g.add((ex.Cade, RDF.type, FOAF.Person))


# OR
EX = Namespace('http://example.org/')


g = Graph()
g = Graph()
g.bind('ex', EX)


ex = Namespace('http://example.org/')
from rdflib.collection import Collection


g.add((ex.Cade, FOAF.name, Literal("Cade", datatype=XSD.string)))
g = Graph()
g.add((ex.Mary, FOAF.name, Literal("Mary", datatype=XSD.string)))
g.bind('ex', EX)
g.add((ex.Cade, RDF.type, FOAF.Person))
g.add((ex.Mary, RDF.type, FOAF.Person))
g.add((ex.Mary, RDF.type, ex.Student))
g.add((ex.Cade, ex.Married, ex.Mary))
g.add((ex.Cade, FOAF.age, Literal('27', datatype=XSD.int)))
g.add((ex.Mary, FOAF.age, Literal('26', datatype=XSD.int)))
g.add((ex.Paris, RDF.type, ex.City))
g.add((ex.France, ex.Capital, ex.Paris))
g.add((ex.Mary, FOAF.interest, ex.hiking))
g.add((ex.Mary, FOAF.interest, ex.Chocolate))
g.add((ex.Mary, FOAF.interest, ex.biology))
g.add((ex.France, ex.City, ex.Paris))
g.add((ex.Mary, ex.characteristic, ex.kind))
g.add((ex.Cade, ex.characteristic, ex.kind))
g.add((ex.France, RDF.type, ex.Country))


donaldTrumpSpouses = BNode()
Collection(g, donaldTrumpSpouses, [
    EX.IvanaTrump, EX.MarlaMaples, EX.MelaniaTrump
])
g.add((EX.Donald_Trump, SCHEMA.spouse, donaldTrumpSpouses))


print(g.serialize(format="turtle"))
print(g.serialize(format='turtle'))
g.serialize(destination='s02_Donald_Trump_spouses_list.ttl', format='turtle')


print(g.serialize(format='turtle'))
</syntaxhighlight>
</syntaxhighlight>


==Basic RDF programming==
=Example lab solutions=


===Different ways to create an address===
==Getting started (Lab 1)==


<syntaxhighlight>
<syntaxhighlight>


from rdflib import Graph, Namespace, URIRef, BNode, Literal
from rdflib import Graph, Namespace
from rdflib.namespace import RDF, FOAF, XSD


g = Graph()
g = Graph()
ex = Namespace("http://example.org/")


ex = Namespace('http://example.org/')


# How to represent the address of Cade Tracey. From probably the worst solution to the best.
g.bind("ex", ex)


# Solution 1 -
#The Mueller Investigation was lead by Robert Mueller.
# Make the entire address into one Literal. However, Generally we want to separate each part of an address into their own triples. This is useful for instance if we want to find only the streets where people live.  
g.add((ex.Mueller_Investigation, ex.leadBy, ex.Robert_Muller))


g.add((ex.Cade_Tracey, ex.livesIn, Literal("1516_Henry_Street, Berkeley, California 94709, USA")))
#It involved Paul Manafort, Rick Gates, George Papadopoulos, Michael Flynn, and Roger Stone.
g.add((ex.Mueller_Investigation, ex.involved, ex.Paul_Manafort))
g.add((ex.Mueller_Investigation, ex.involved, ex.Rick_Gates))
g.add((ex.Mueller_Investigation, ex.involved, ex.George_Papadopoulos))
g.add((ex.Mueller_Investigation, ex.involved, ex.Michael_Flynn))
g.add((ex.Mueller_Investigation, ex.involved, ex.Michael_Cohen))
g.add((ex.Mueller_Investigation, ex.involved, ex.Roger_Stone))


# --- Paul Manafort ---
#Paul Manafort was business partner of Rick Gates.
g.add((ex.Paul_Manafort, ex.businessManager, ex.Rick_Gates))
# He was campaign chairman for Trump
g.add((ex.Paul_Manafort, ex.campaignChairman, ex.Donald_Trump))


# Solution 2 -
# He was charged with money laundering, tax evasion, and foreign lobbying.
# Seperate the different pieces information into their own triples
g.add((ex.Paul_Manafort, ex.chargedWith, ex.MoneyLaundering))
g.add((ex.Paul_Manafort, ex.chargedWith, ex.TaxEvasion))
g.add((ex.Paul_Manafort, ex.chargedWith, ex.ForeignLobbying))


g.add((ex.Cade_tracey, ex.street, Literal("1516_Henry_Street")))
# He was convicted for bank and tax fraud.
g.add((ex.Cade_tracey, ex.city, Literal("Berkeley")))
g.add((ex.Paul_Manafort, ex.convictedFor, ex.BankFraud))
g.add((ex.Cade_tracey, ex.state, Literal("California")))
g.add((ex.Paul_Manafort, ex.convictedFor, ex.TaxFraud))
g.add((ex.Cade_tracey, ex.zipcode, Literal("94709")))
g.add((ex.Cade_tracey, ex.country, Literal("USA")))


# He pleaded guilty to conspiracy.
g.add((ex.Paul_Manafort, ex.pleadGuiltyTo, ex.Conspiracy))
# He was sentenced to prison.
g.add((ex.Paul_Manafort, ex.sentencedTo, ex.Prison))
# He negotiated a plea agreement.
g.add((ex.Paul_Manafort, ex.negoiated, ex.PleaBargain))


# Solution 3 - Some parts of the addresses can make more sense to be resources than Literals.
# --- Rick Gates ---
# Larger concepts like a city or state are typically represented as resources rather than Literals, but this is not necesarilly a requirement in the case that you don't intend to say more about them.  
#Rick Gates was charged with money laundering, tax evasion and foreign lobbying.
g.add((ex.Rick_Gates, ex.chargedWith, ex.MoneyLaundering))
g.add((ex.Rick_Gates, ex.chargedWith, ex.TaxEvasion))
g.add((ex.Rick_Gates, ex.chargedWith, ex.ForeignLobbying))


g.add((ex.Cade_tracey, ex.street, Literal("1516_Henry_Street")))
#He pleaded guilty to conspiracy and lying to FBI.
g.add((ex.Cade_tracey, ex.city, ex.Berkeley))
g.add((ex.Rick_Gates, ex.pleadGuiltyTo, ex.Conspiracy))
g.add((ex.Cade_tracey, ex.state, ex.California))
g.add((ex.Rick_Gates, ex.pleadGuiltyTo, ex.LyingToFBI))
g.add((ex.Cade_tracey, ex.zipcode, Literal("94709")))
g.add((ex.Cade_tracey, ex.country, ex.USA))


#Use the serialize method to write out the model in different formats on screen
print(g.serialize(format="ttl"))
# g.serialize("lab1.ttl", format="ttl") #or to file


# Solution 4
#Loop through the triples in the model to print out all triples that have pleading guilty as predicate
# Grouping of the information into an Address. We can Represent the address concept with its own URI OR with a Blank Node.
for subject, object in g[ : ex.pleadGuiltyTo : ]:
# One advantage of this is that we can easily remove the entire address, instead of removing each individual part of the address.  
    print(subject, ex.pleadGuiltyTo, object)
# Solution 4 or 5 is how I would recommend to make addresses. Here, ex.CadeAddress could also be called something like ex.address1 or so on, if you want to give each address a unique ID.


# Address URI - CadeAdress
# Michael Cohen, Michael Flynn and the lying is part of lab 2 and therefore the answer is not provided this week


g.add((ex.Cade_Tracey, ex.address, ex.CadeAddress))
#Write a method (function) that submits your model for rendering and saves the returned image to file.
g.add((ex.CadeAddress, RDF.type, ex.Address))
import requests
g.add((ex.CadeAddress, ex.street, Literal("1516 Henry Street")))
import shutil
g.add((ex.CadeAddress, ex.city, ex.Berkeley))
g.add((ex.CadeAddress, ex.state, ex.California))
g.add((ex.CadeAddress, ex.postalCode, Literal("94709")))
g.add((ex.CadeAddress, ex.country, ex.USA))


# OR
def graphToImage(graph):
    data = {"rdf":graph, "from":"ttl", "to":"png"}
    link = "http://www.ldf.fi/service/rdf-grapher"
    response = requests.get(link, params = data, stream=True)
    # print(response.content)
    print(response.raw)
    with open("lab1.png", "wb") as fil:
        shutil.copyfileobj(response.raw, fil)


# Blank node for Address. 
graph = g.serialize(format="ttl")
address = BNode()
graphToImage(graph)
g.add((ex.Cade_Tracey, ex.address, address))
</syntaxhighlight>
g.add((address, RDF.type, ex.Address))
g.add((address, ex.street, Literal("1516 Henry Street", datatype=XSD.string)))
g.add((address, ex.city, ex.Berkeley))
g.add((address, ex.state, ex.California))
g.add((address, ex.postalCode, Literal("94709", datatype=XSD.string)))
g.add((address, ex.country, ex.USA))


==RDF programming with RDFlib (Lab 2)==


# Solution 5 using existing vocabularies for address
<syntaxhighlight>


# (in this case https://schema.org/PostalAddress from schema.org).
from rdflib import Graph, URIRef, Namespace, Literal, XSD, BNode
# Also using existing ontology for places like California. (like http://dbpedia.org/resource/California from dbpedia.org)
from rdflib.collection import Collection


schema = Namespace("https://schema.org/")
g = Graph()
dbp = Namespace("https://dpbedia.org/resource/")
g.parse("lab1.ttl", format="ttl") #Retrives the triples from lab 1


g.add((ex.Cade_Tracey, schema.address, ex.CadeAddress))
ex = Namespace('http://example.org/')
g.add((ex.CadeAddress, RDF.type, schema.PostalAddress))
g.add((ex.CadeAddress, schema.streetAddress, Literal("1516 Henry Street")))
g.add((ex.CadeAddress, schema.addresCity, dbp.Berkeley))
g.add((ex.CadeAddress, schema.addressRegion, dbp.California))
g.add((ex.CadeAddress, schema.postalCode, Literal("94709")))
g.add((ex.CadeAddress, schema.addressCountry, dbp.United_States))


</syntaxhighlight>
# --- Michael Cohen ---
#Michael Cohen was Donald Trump's attorney.
g.add((ex.Michael_Cohen, ex.attorneyTo, ex.Donald_Trump))
#He pleaded guilty to lying to the FBI.
g.add((ex.Michael_Cohen, ex.pleadGuiltyTo, ex.LyingToCongress))


===Typed Literals===
# --- Michael Flynn ---
<syntaxhighlight>
#Michael Flynn was adviser to Trump.
from rdflib import Graph, Literal, Namespace
g.add((ex.Michael_Flynn, ex.adviserTo, ex.Donald_Trump))
from rdflib.namespace import XSD
#He pleaded guilty to lying to the FBI.
g = Graph()
g.add((ex.Michael_Flynn, ex.pleadGuiltyTo, ex.LyingToFBI))
ex = Namespace("http://example.org/")
# He negotiated a plea agreement.
g.add((ex.Michael_Flynn, ex.negoiated, ex.PleaBargain))  


g.add((ex.Cade, ex.age, Literal(27, datatype=XSD.integer)))
#How can you modify your knowledge graph to account for the different lying?
g.add((ex.Cade, ex.gpa, Literal(3.3, datatype=XSD.float)))
#Remove these to not have duplicates
g.add((ex.Cade, FOAF.name, Literal("Cade Tracey", datatype=XSD.string)))
g.remove((ex.Michael_Flynn, ex.pleadGuiltyTo, ex.LyingToFBI))
g.add((ex.Cade, ex.birthday, Literal("2006-01-01", datatype=XSD.date)))
g.remove((ex.Michael_Flynn, ex.negoiated, ex.PleaBargain))
</syntaxhighlight>
g.remove((ex.Rick_Gates, ex.pleadGuiltyTo, ex.LyingToFBI))
g.remove((ex.Rick_Gates, ex.pleadGuiltyTo, ex.Conspiracy))
g.remove((ex.Rick_Gates, ex.chargedWith, ex.ForeignLobbying))
g.remove((ex.Rick_Gates, ex.chargedWith, ex.MoneyLaundering))
g.remove((ex.Rick_Gates, ex.chargedWith, ex.TaxEvasion))
g.remove((ex.Michael_Cohen, ex.pleadGuiltyTo, ex.LyingToCongress))


# --- Michael Flynn ---
FlynnLying = BNode()
g.add((FlynnLying, ex.crime, ex.LyingToFBI))
g.add((FlynnLying, ex.pleadGulityOn, Literal("2017-12-1", datatype=XSD.date)))
g.add((FlynnLying, ex.liedAbout, Literal("His communications with a former Russian ambassador during the presidential transition", datatype=XSD.string)))
g.add((FlynnLying, ex.pleaBargain, Literal("true", datatype=XSD.boolean)))
g.add((ex.Michael_Flynn, ex.pleadGuiltyTo, FlynnLying))


===Writing and reading graphs/files===
# --- Rick Gates ---
GatesLying = BNode()
Crimes = BNode()
Charged = BNode()
Collection(g, Crimes, [ex.LyingToFBI, ex.Conspiracy])
Collection(g, Charged, [ex.ForeignLobbying, ex.MoneyLaundering, ex.TaxEvasion])
g.add((GatesLying, ex.crime, Crimes))
g.add((GatesLying, ex.chargedWith, Charged))
g.add((GatesLying, ex.pleadGulityOn, Literal("2018-02-23", datatype=XSD.date)))
g.add((GatesLying, ex.pleaBargain, Literal("true", datatype=XSD.boolean)))
g.add((ex.Rick_Gates, ex.pleadGuiltyTo, GatesLying))


<syntaxhighlight>
# --- Michael Cohen ---
  # Writing the graph to a file on your system. Possible formats = turtle, n3, xml, nt.
CohenLying = BNode()
g.serialize(destination="triples.txt", format="turtle")
g.add((CohenLying, ex.crime, ex.LyingToCongress))
g.add((CohenLying, ex.liedAbout, ex.TrumpRealEstateDeal))
g.add((CohenLying, ex.prosecutorsAlleged, Literal("In an August 2017 letter Cohen sent to congressional committees investigating Russian election interference, he falsely stated that the project ended in January 2016", datatype=XSD.string)))
g.add((CohenLying, ex.mullerInvestigationAlleged, Literal("Cohen falsely stated that he had never agreed to travel to Russia for the real estate deal and that he did not recall any contact with the Russian government about the project", datatype=XSD.string)))
g.add((CohenLying, ex.pleadGulityOn, Literal("2018-11-29", datatype=XSD.date)))
g.add((CohenLying, ex.pleaBargain, Literal("true", datatype=XSD.boolean)))
g.add((ex.Michael_Cohen, ex.pleadGuiltyTo, CohenLying))


  # Parsing a local file
print(g.serialize(format="ttl"))
parsed_graph = g.parse(location="triples.txt", format="turtle")


  # Parsing a remote endpoint like Dbpedia
#Save (serialize) your graph to a Turtle file.
dbpedia_graph = g.parse("http://dbpedia.org/resource/Pluto")
# g.serialize("lab2.ttl", format="ttl")
</syntaxhighlight>


===Graph Binding===
#Add a few triples to the Turtle file with more information about Donald Trump.
<syntaxhighlight>
'''
#Graph Binding is useful for at least two reasons:
ex:Donald_Trump ex:address [ ex:city ex:Palm_Beach ;
#(1) We no longer need to specify prefixes with SPARQL queries if they are already binded to the graph.
            ex:country ex:United_States ;
#(2) When serializing the graph, the serialization will show the correct expected prefix
            ex:postalCode 33480 ;
# instead of default namespace names ns1, ns2 etc.
            ex:residence ex:Mar_a_Lago ;
            ex:state ex:Florida ;
            ex:streetName "1100 S Ocean Blvd"^^xsd:string ] ;
    ex:previousAddress [ ex:city ex:Washington_DC ;
            ex:country ex:United_States ;
            ex:phoneNumber "1 202 456 1414"^^xsd:integer ;
            ex:postalCode "20500"^^xsd:integer ;
            ex:residence ex:The_White_House ;
            ex:streetName "1600 Pennsylvania Ave."^^xsd:string ];
    ex:marriedTo ex:Melania_Trump;
    ex:fatherTo (ex:Ivanka_Trump ex:Donald_Trump_Jr ex: ex:Tiffany_Trump ex:Eric_Trump ex:Barron_Trump).
'''


g = Graph()
#Read (parse) the Turtle file back into a Python program, and check that the new triples are there
def serialize_Graph():
    newGraph = Graph()
    newGraph.parse("lab2.ttl")
    print(newGraph.serialize())


ex = Namespace("http://example.org/")
# serialize_Graph() #Don't need this to run until after adding the triples above to the ttl file
dbp = Namespace("http://dbpedia.org/resource/")
schema = Namespace("https://schema.org/")


g.bind("ex", ex)
#Write a method (function) that starts with Donald Trump prints out a graph depth-first to show how the other graph nodes are connected to him
g.bind("dbp", dbp)
visited_nodes = set()
g.bind("schema", schema)
</syntaxhighlight>


===Collection Example===
def create_Tree(model, nodes):
    #Traverse the model breadth-first to create the tree.
    global visited_nodes
    tree = Graph()
    children = set()
    visited_nodes |= set(nodes)
    for s, p, o in model:
        if s in nodes and o not in visited_nodes:
            tree.add((s, p, o))
            visited_nodes.add(o)
            children.add(o)
        if o in nodes and s not in visited_nodes:
            invp = URIRef(f'{p}_inv') #_inv represents inverse of
            tree.add((o, invp, s))
            visited_nodes.add(s)
            children.add(s)
    if len(children) > 0:
        children_tree = create_Tree(model, children)
        for triple in children_tree:
            tree.add(triple)
    return tree


<syntaxhighlight>
def print_Tree(tree, root, indent=0):
from rdflib import Graph, Namespace
    #Print the tree depth-first.
from rdflib.collection import Collection
    print(str(root))
    for s, p, o in tree:
        if s==root:
            print('    '*indent + '  ' + str(p), end=' ')
            print_Tree(tree, o, indent+1)
   
tree = create_Tree(g, [ex.Donald_Trump])
print_Tree(tree, ex.Donald_Trump)


</syntaxhighlight>


# Sometimes we want to add many objects or subjects for the same predicate at once.
==SPARQL Programming (Lab 4)==
# In these cases we can use Collection() to save some time.
'''NOTE: These tasks were performed on the old dataset, with the new dataset, some of these answers would be different.'''
# In this case I want to add all countries that Emma has visited at once.
<syntaxhighlight>


b = BNode()
from rdflib import Graph, Namespace, RDF, FOAF
g.add((ex.Emma, ex.visit, b))
from SPARQLWrapper import SPARQLWrapper, JSON, POST, GET, TURTLE
Collection(g, b,
    [ex.Portugal, ex.Italy, ex.France, ex.Germany, ex.Denmark, ex.Sweden])


# OR
g = Graph()
g.parse("Russia_investigation_kg.ttl")


g.add((ex.Emma, ex.visit, ex.EmmaVisits))
# ----- RDFLIB -----
Collection(g, ex.EmmaVisits,
ex = Namespace('http://example.org#')
    [ex.Portugal, ex.Italy, ex.France, ex.Germany, ex.Denmark, ex.Sweden])


</syntaxhighlight>
NS = {
    '': ex,
    'rdf': RDF,
    'foaf': FOAF,
}


==SPARQL==
# Print out a list of all the predicates used in your graph.
task1 = g.query("""
SELECT DISTINCT ?p WHERE{
    ?s ?p ?o .
}
""", initNs=NS)


Also see the [[SPARQL Examples]] page!
print(list(task1))


===Querying a local ("in memory") graph===
# Print out a sorted list of all the presidents represented in your graph.
task2 = g.query("""
SELECT DISTINCT ?president WHERE{
    ?s :president ?president .
}
ORDER BY ?president
""", initNs=NS)


Example contents of the file family.ttl:
print(list(task2))
@prefix rex: <http://example.org/royal#> .
@prefix fam: <http://example.org/family#> .
rex:IngridAlexandra fam:hasParent rex:HaakonMagnus .
rex:SverreMagnus fam:hasParent rex:HaakonMagnus .
rex:HaakonMagnus fam:hasParent rex:Harald .
rex:MarthaLouise fam:hasParent rex:Harald .
rex:HaakonMagnus fam:hasSister rex:MarthaLouise .


import rdflib
# Create dictionary (Python dict) with all the represented presidents as keys. For each key, the value is a list of names of people indicted under that president.
task3_dic = {}
g = rdflib.Graph()
g.parse("family.ttl", format='ttl')
qres = g.query("""
PREFIX fam: <http://example.org/family#>
    SELECT ?child ?sister WHERE {
        ?child fam:hasParent ?parent .
        ?parent fam:hasSister ?sister .
    }""")
for row in qres:
    print("%s has aunt %s" % row)


With a prepared query, you can write the query once, and then bind some of the variables each time you use it:
task3 = g.query("""
import rdflib
SELECT ?president ?person WHERE{
    ?s :president ?president;
g = rdflib.Graph()
      :name ?person;
g.parse("family.ttl", format='ttl')
      :outcome :indictment.
}
q = rdflib.plugins.sparql.prepareQuery(
""", initNs=NS)
        """SELECT ?child ?sister WHERE {
                  ?child fam:hasParent ?parent .
                  ?parent fam:hasSister ?sister .
        }""",
        initNs = { "fam": "http://example.org/family#"})
sm = rdflib.URIRef("http://example.org/royal#SverreMagnus")
for row in g.query(q, initBindings={'child': sm}):
        print(row)


===Select all contents of lists (rdfllib.Collection)===
for president, person in task3:
<syntaxhighlight>
    if president not in task3_dic:
        task3_dic[president] = [person]
    else:
        task3_dic[president].append(person)


# rdflib.Collection has a different interntal structure so it requires a slightly more advance query. Here I am selecting all places that Emma has visited.
print(task3_dic)


PREFIX ex:  <http://example.org/>
# Use an ASK query to investigate whether Donald Trump has pardoned more than 5 people.
PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>


SELECT ?visit
# This task is a lot trickier than it needs to be. As far as I'm aware RDFLib has no HAVING support, so a query like this:
WHERE {
task4 = g.query("""
  ex:Emma ex:visit/rdf:rest*/rdf:first ?visit
ASK {
  SELECT (COUNT(?s) as ?count) WHERE{
    ?s :pardoned :true;
    :president :Bill_Clinton  .
    }
    HAVING (?count > 5)
}
}
</syntaxhighlight>
""", initNs=NS)


print(task4.askAnswer)


===Using parameters/variables in rdflib queries===
# Which works fine in Blazegraph and is a valid SPARQL query will always provide false in RDFLib, cause it uses HAVING. Instead you have to use a nested SELECT query like below, where you use FILTER instead of HAVING. Donald Trump has no pardons, so I have instead chosen Bill Clinton (which has 13 pardons) to check if the query works.


<syntaxhighlight>
task4 = g.query("""
from rdflib import Graph, Namespace, URIRef
    ASK{
from rdflib.plugins.sparql import prepareQuery
        SELECT ?count WHERE{{
          SELECT (COUNT(?s) as ?count) WHERE{
            ?s :pardoned :true;
                  :president :Bill_Clinton  .
                }}
        FILTER (?count > 5)
        }
    }
""", initNs=NS)


g = Graph()
print(task4.askAnswer)
ex = Namespace("http://example.org/")
g.bind("ex", ex)


g.add((ex.Cade, ex.livesIn, ex.France))
# Use a DESCRIBE query to create a new graph with information about Donald Trump. Print out the graph in Turtle format.
g.add((ex.Anne, ex.livesIn, ex.Norway))
g.add((ex.Sofie, ex.livesIn, ex.Sweden))
g.add((ex.Per, ex.livesIn, ex.Norway))
g.add((ex.John, ex.livesIn, ex.USA))


# By all accounts, it seems DESCRIBE queries are yet to be implemented in RDFLib, but they are attempting to implement it: https://github.com/RDFLib/rdflib/pull/2221 (Issue and proposed solution raised) & https://github.com/RDFLib/rdflib/commit/2325b4a81724c1ccee3a131067db4fbf9b4e2629 (Solution committed to RDFLib). This solution does not work. However, this proposed solution should work if DESCRIBE is implemented in RDFLib


def find_people_from_country(country):
# task5 = g.query("""  
        country = URIRef(ex + country)
# DESCRIBE :Donald_Trump
        q = prepareQuery(
# """, initNs=NS)
        """
        PREFIX ex: <http://example.org/>
        SELECT ?person WHERE {
        ?person ex:livesIn ?country.
        }
        """)


        capital_result = g.query(q, initBindings={'country': country})
# print(task5.serialize())


        for row in capital_result:
# ----- SPARQLWrapper -----
            print(row)


find_people_from_country("Norway")
namespace = "kb" #Default namespace
</syntaxhighlight>
sparql = SPARQLWrapper("http://localhost:9999/blazegraph/namespace/"+ namespace + "/sparql") #Replace localhost:9999 with your URL


===SELECTING data from Blazegraph via Python===
# The current dates are URIs, we would want to change them to Literals with datatype "date" for task 1 & 2
<syntaxhighlight>
update_str = """
    PREFIX ns1: <http://example.org#>


from SPARQLWrapper import SPARQLWrapper, JSON
    DELETE {
        ?s ns1:cp_date ?cp;
            ns1:investigation_end ?end;
            ns1:investigation_start ?start.
    }
    INSERT{
        ?s ns1:cp_date ?cpDate;
            ns1:investigation_end ?endDate;
            ns1:investigation_start ?startDate.
    }
    WHERE{
        ?s ns1:cp_date ?cp . #Date conviction was recieved
        BIND (replace(str(?cp), str(ns1:), "")  AS ?cpRemoved)
        BIND (STRDT(STR(?cpRemoved), xsd:date) AS ?cpDate)
       
        ?s ns1:investigation_end ?end . #Investigation End
        BIND (replace(str(?end), str(ns1:), "")  AS ?endRemoved)
        BIND (STRDT(STR(?endRemoved), xsd:date) AS ?endDate)
       
        ?s ns1:investigation_start ?start . #Investigation Start
        BIND (replace(str(?start), str(ns1:), "")  AS ?startRemoved)
        BIND (STRDT(STR(?startRemoved), xsd:date) AS ?startDate)
}"""


# This creates a server connection to the same URL that contains the graphic interface for Blazegraph.  
sparql.setQuery(update_str)
# You also need to add "sparql" to end of the URL like below.
sparql.setMethod(POST)
 
sparql.query()
sparql = SPARQLWrapper("http://localhost:9999/blazegraph/sparql")
 
# SELECT all triples in the database.


# Ask whether there was an ongoing indictment on the date 1990-01-01.
sparql.setQuery("""
sparql.setQuery("""
     SELECT DISTINCT ?p WHERE {
     PREFIX ns1: <http://example.org#>
    ?s ?p ?o.
    ASK {
        SELECT ?end ?start
        WHERE{
            ?s ns1:investigation_end ?end;
              ns1:investigation_start ?start;
              ns1:outcome ns1:indictment.
            FILTER(?start <= "1990-01-01"^^xsd:date && ?end >= "1990-01-01"^^xsd:date)
    }
     }
     }
""")
""")
sparql.setReturnFormat(JSON)
sparql.setReturnFormat(JSON)
results = sparql.query().convert()
results = sparql.query().convert()
print(f"Are there any investigation on the 1990-01-01: {results['boolean']}")


for result in results["results"]["bindings"]:
# List ongoing indictments on that date 1990-01-01.
    print(result["p"]["value"])
 
# SELECT all interests of Cade
 
sparql.setQuery("""
sparql.setQuery("""
     PREFIX ex: <http://example.org/>
     PREFIX ns1: <http://example.org#>
     SELECT DISTINCT ?interest WHERE {
     SELECT ?s
    ex:Cade ex:interest ?interest.
    WHERE{
        ?s ns1:investigation_end ?end;
          ns1:investigation_start ?start;
          ns1:outcome ns1:indictment.
        FILTER(?start <= "1990-01-01"^^xsd:date && ?end >= "1990-01-01"^^xsd:date)
     }
     }
""")
""")
sparql.setReturnFormat(JSON)
sparql.setReturnFormat(JSON)
results = sparql.query().convert()
results = sparql.query().convert()


print("The ongoing investigations on the 1990-01-01 are:")
for result in results["results"]["bindings"]:
for result in results["results"]["bindings"]:
     print(result["interest"]["value"])
     print(result["s"]["value"])
</syntaxhighlight>


===Updating data from Blazegraph via Python===
# Describe investigation number 100 (muellerkg:investigation_100).
<syntaxhighlight>
from SPARQLWrapper import SPARQLWrapper, POST, DIGEST
 
namespace = "kb"
sparql = SPARQLWrapper("http://localhost:9999/blazegraph/namespace/"+ namespace + "/sparql")
 
sparql.setMethod(POST)
sparql.setQuery("""
sparql.setQuery("""
     PREFIX ex: <http://example.org/>
     PREFIX ns1: <http://example.org#>
     INSERT DATA{
     DESCRIBE ns1:investigation_100
    ex:Cade ex:interest ex:Mathematics.
    }
""")
""")


results = sparql.query()
sparql.setReturnFormat(TURTLE)
print(results.response.read())
results = sparql.query().convert()


print(results.serialize())


</syntaxhighlight>
# Print out a list of all the types used in your graph.
===Retrieving data from Wikidata with SparqlWrapper===
sparql.setQuery("""
<syntaxhighlight>
    PREFIX ns1: <http://example.org#>
from SPARQLWrapper import SPARQLWrapper, JSON
    PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>


sparql = SPARQLWrapper("https://query.wikidata.org/sparql")
     SELECT DISTINCT ?types
# In the query I want to select all the Vitamins in wikidata.
    WHERE{
 
        ?s rdf:type ?types .  
sparql.setQuery("""
    }
     SELECT ?nutrient ?nutrientLabel WHERE
{
  ?nutrient wdt:P279 wd:Q34956.
  SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
}
""")
""")


sparql.setReturnFormat(JSON)
sparql.setReturnFormat(JSON)
results = sparql.query().convert()
results = sparql.query().convert()
rdf_Types = []


for result in results["results"]["bindings"]:
for result in results["results"]["bindings"]:
     print(result["nutrient"]["value"], "  ", result["nutrientLabel"]["value"])
     rdf_Types.append(result["types"]["value"])
</syntaxhighlight>


print(rdf_Types)


More examples can be found in the example section on the official query service here: https://query.wikidata.org/.
# Update the graph to that every resource that is an object in a muellerkg:investigation triple has the rdf:type muellerkg:Investigation.
update_str = """
    PREFIX ns1: <http://example.org#>
    PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>


===Download from BlazeGraph===
    INSERT{
        ?invest rdf:type ns1:Investigation .
    }
    WHERE{
        ?s ns1:investigation ?invest .
}"""


<syntaxhighlight>
sparql.setQuery(update_str)
"""
sparql.setMethod(POST)
Dumps a database to a local RDF file.
sparql.query()
You need to install the SPARQLWrapper package first...
"""


import datetime
#To Test
from SPARQLWrapper import SPARQLWrapper, RDFXML
sparql.setQuery("""
    prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
    PREFIX ns1: <http://example.org#>


# your namespace, the default is 'kb'
    ASK{
ns = 'kb'
        ns1:watergate rdf:type ns1:Investigation.
    }
""")


# the SPARQL endpoint
sparql.setReturnFormat(JSON)
endpoint = 'http://info216.i2s.uib.no/bigdata/namespace/' + ns + '/sparql'
results = sparql.query().convert()
print(results['boolean'])


# - the endpoint just moved, the old one was:
# Update the graph to that every resource that is an object in a muellerkg:person triple has the rdf:type muellerkg:IndictedPerson.
# endpoint = 'http://i2s.uib.no:8888/bigdata/namespace/' + ns + '/sparql'
update_str = """
    PREFIX ns1: <http://example.org#>
    PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>


# create wrapper
    INSERT{
wrapper = SPARQLWrapper(endpoint)
        ?person rdf:type ns1:IndictedPerson .
    }
    WHERE{
        ?s ns1:person ?person .
}"""


# prepare the SPARQL update
sparql.setQuery(update_str)
wrapper.setQuery('CONSTRUCT { ?s ?p ?o } WHERE { ?s ?p ?o }')
sparql.setMethod(POST)
wrapper.setReturnFormat(RDFXML)
sparql.query()


# execute the SPARQL update and convert the result to an rdflib.Graph
#To test, run the query in the above task, replacing the ask query with e.g. ns1:Deborah_Gore_Dean rdf:type ns1:IndictedPerson
graph = wrapper.query().convert()


# the destination file, with code to make it timestamped
# Update the graph so all the investigation nodes (such as muellerkg:watergate) become the subject in a dc:title triple with the corresponding string (watergate) as the literal.
destfile = 'rdf_dumps/slr-kg4news-' + datetime.datetime.now().strftime('%Y%m%d-%H%M') + '.rdf'
update_str = """
    PREFIX ns1: <http://example.org#>
    PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
    PREFIX dc: <http://purl.org/dc/elements/1.1/>


# serialize the result to file
    INSERT{
graph.serialize(destination=destfile, format='ttl')
        ?invest dc:title ?investString.
    }
    WHERE{
        ?s ns1:investigation ?invest .
        BIND (replace(str(?invest), str(ns1:), "") AS ?investString)
}"""


# report and quit
sparql.setQuery(update_str)
print('Wrote %u triples to file %s .' %
sparql.setMethod(POST)
      (len(res), destfile))
sparql.query()
</syntaxhighlight>


===Query Dbpedia with SparqlWrapper===
#Same test as above, replace it with e.g. ns1:watergate dc:title "watergate"


<syntaxhighlight>
# Print out a sorted list of all the indicted persons represented in your graph.
from SPARQLWrapper import SPARQLWrapper, JSON
sparql.setQuery("""
 
    PREFIX ns1: <http://example.org#>
sparql = SPARQLWrapper("http://dbpedia.org/sparql")
    PREFIX foaf: <http://xmlns.com/foaf/0.1/>


sparql.setQuery("""
     SELECT ?name
    PREFIX dbr: <http://dbpedia.org/resource/>
     WHERE{
    PREFIX dbo: <http://dbpedia.org/ontology/>
     ?s  ns1:person ?name;
    PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
        ns1:outcome ns1:indictment.
     SELECT ?comment
     WHERE {
     dbr:Barack_Obama rdfs:comment ?comment.
    FILTER (langMatches(lang(?comment),"en"))
     }
     }
    ORDER BY ?name
""")
""")


sparql.setReturnFormat(JSON)
sparql.setReturnFormat(JSON)
results = sparql.query().convert()
results = sparql.query().convert()
names = []


for result in results["results"]["bindings"]:
for result in results["results"]["bindings"]:
     print(result["comment"]["value"])
     names.append(result["name"]["value"])
</syntaxhighlight>


==Lifting CSV to RDF==
print(names)


<syntaxhighlight>
# Print out the minimum, average and maximum indictment days for all the indictments in the graph.
from rdflib import Graph, Literal, Namespace, URIRef
sparql.setQuery("""
from rdflib.namespace import RDF, FOAF, RDFS, OWL
    prefix xsd: <http://www.w3.org/2001/XMLSchema#>
import pandas as pd
    PREFIX ns1: <http://example.org#>


g = Graph()
    SELECT (AVG(?daysRemoved) as ?avg) (MAX(?daysRemoved) as ?max) (MIN(?daysRemoved) as ?min) WHERE{
ex = Namespace("http://example.org/")
        ?s  ns1:indictment_days ?days;
g.bind("ex", ex)
            ns1:outcome ns1:indictment.
   
    BIND (replace(str(?days), str(ns1:), "")  AS ?daysR)
    BIND (STRDT(STR(?daysR), xsd:float) AS ?daysRemoved)
}
""")


# Load the CSV data as a pandas Dataframe.
sparql.setReturnFormat(JSON)
csv_data = pd.read_csv("task1.csv")
results = sparql.query().convert()


# Here I deal with spaces (" ") in the data. I replace them with "_" so that URI's become valid.
for result in results["results"]["bindings"]:
csv_data = csv_data.replace(to_replace=" ", value="_", regex=True)
    print(f'The longest an investigation lasted was: {result["max"]["value"]}')
    print(f'The shortest an investigation lasted was: {result["min"]["value"]}')
    print(f'The average investigation lasted: {result["avg"]["value"]}')


# Here I mark all missing/empty data as "unknown". This makes it easy to delete triples containing this later.
# Print out the minimum, average and maximum indictment days for all the indictments in the graph per investigation.
csv_data = csv_data.fillna("unknown")
sparql.setQuery("""
    prefix xsd: <http://www.w3.org/2001/XMLSchema#>
    PREFIX ns1: <http://example.org#>


# Loop through the CSV data, and then make RDF triples.
    SELECT ?investigation (AVG(?daysRemoved) as ?avg) (MAX(?daysRemoved) as ?max) (MIN(?daysRemoved) as ?min)  WHERE{
for index, row in csv_data.iterrows():
     ?s  ns1:indictment_days ?days;
     # The names of the people act as subjects.
        ns1:outcome ns1:indictment;
    subject = row['Name']
        ns1:investigation ?investigation.
    # Create triples: e.g. "Cade_Tracey - age - 27"
      
     g.add((URIRef(ex + subject), URIRef(ex + "age"), Literal(row["Age"])))
    BIND (replace(str(?days), str(ns1:), "") AS ?daysR)
     g.add((URIRef(ex + subject), URIRef(ex + "married"), URIRef(ex + row["Spouse"])))
     BIND (STRDT(STR(?daysR), xsd:float) AS ?daysRemoved)
     g.add((URIRef(ex + subject), URIRef(ex + "country"), URIRef(ex + row["Country"])))
     }
    GROUP BY ?investigation
""")


    # If We want can add additional RDF/RDFS/OWL information e.g
sparql.setReturnFormat(JSON)
    g.add((URIRef(ex + subject), RDF.type, FOAF.Person))
results = sparql.query().convert()


# I remove triples that I marked as unknown earlier.
for result in results["results"]["bindings"]:
g.remove((None, None, URIRef("http://example.org/unknown")))
    print(f'{result["investigation"]["value"]} - min: {result["min"]["value"]}, max: {result["max"]["value"]}, avg: {result["avg"]["value"]}')
 
# Clean printing of the graph.
print(g.serialize(format="turtle").decode())
</syntaxhighlight>
 
===CSV file for above example===


<syntaxhighlight>
"Name","Age","Spouse","Country"
"Cade Tracey","26","Mary Jackson","US"
"Bob Johnson","21","","Canada"
"Mary Jackson","25","","France"
"Phil Philips","32","Catherine Smith","Japan"
</syntaxhighlight>
</syntaxhighlight>


 
==CSV To RDF (Lab 5)==
=Coding Tasks Lab 6=
<syntaxhighlight>
<syntaxhighlight>
import pandas as pd


#Imports
import re
from pandas import *
from numpy import nan
from rdflib import Graph, Namespace, URIRef, Literal, RDF, XSD, FOAF
from spotlight import SpotlightException, annotate


from rdflib import Graph, Namespace, URIRef, Literal, BNode
SERVER = "https://api.dbpedia-spotlight.org/en/annotate"
from rdflib.namespace import RDF, XSD
# Test around with the confidence, and see how many names changes depending on the confidence. However, be aware that anything lower than this (0.83) it will replace James W. McCord and other names that includes James with LeBron James
CONFIDENCE = 0.83


def annotate_entity(entity, filters={'types': 'DBpedia:Person'}):
annotations = []
try:
annotations = annotate(address=SERVER, text=entity, confidence=CONFIDENCE, filters=filters)
except SpotlightException as e:
print(e)
return annotations


g = Graph()
ex = Namespace("http://example.org/")
ex = Namespace("http://example.org/")
sem = Namespace("http://semanticweb.cs.vu.nl/2009/11/sem/")
g = Graph()
g.bind("ex", ex)
g.bind("sem", sem)
# Removing unwanted characters
df = pd.read_csv('russia-investigation.csv')
# Here I deal with spaces (" ") in the data. I replace them with "_" so that URI's become valid.
df = df.replace(to_replace=" ", value="_", regex=True)
# This may seem odd, but in the data set we have a name like this:("Scooter"). So we have to remove quotation marks
df = df.replace(to_replace=f'"', value="", regex=True)
# # Here I mark all missing/empty data as "unknown". This makes it easy to delete triples containing this later.
df = df.fillna("unknown")
# Loop through the CSV data, and then make RDF triples.
for index, row in df.iterrows():
    name = row['investigation']
    investigation = URIRef(ex + name)
    g.add((investigation, RDF.type, sem.Event))
    investigation_start = row["investigation-start"]
    g.add((investigation, sem.hasBeginTimeStamp, Literal(
        investigation_start, datatype=XSD.datetime)))
    investigation_end = row["investigation-end"]
    g.add((investigation, sem.hasEndTimeStamp, Literal(
        investigation_end, datatype=XSD.datetime)))
    investigation_end = row["investigation-days"]
    g.add((investigation, sem.hasXSDDuration, Literal(
        investigation_end, datatype=XSD.Days)))
    person = row["name"]
    person = URIRef(ex + person)
    g.add((investigation, sem.Actor, person))
    result = row['type']
    g.add((investigation, sem.hasSubEvent, Literal(result, datatype=XSD.string)))
    overturned = row["overturned"]
    g.add((investigation, ex.overtuned, Literal(overturned, datatype=XSD.boolean)))
    pardoned = row["pardoned"]
    g.add((investigation, ex.pardon, Literal(pardoned, datatype=XSD.boolean)))
g.serialize("output.ttl", format="ttl")
print(g.serialize(format="turtle"))
</syntaxhighlight>
<!--
==Lifting XML to RDF==
<syntaxhighlight>
from rdflib import Graph, Literal, Namespace, URIRef
from rdflib.namespace import RDF, XSD, RDFS
import xml.etree.ElementTree as ET
g = Graph()
ex = Namespace("http://example.org/TV/")
prov = Namespace("http://www.w3.org/ns/prov#")
g.bind("ex", ex)
g.bind("ex", ex)
g.bind("prov", prov)


tree = ET.parse("tv_shows.xml")
#Pandas' read_csv function to load russia-investigation.csv
root = tree.getroot()
df = read_csv("russia-investigation.csv")
#Replaces all instances of nan to None type with numpy's nan
df = df.replace(nan, None)


for tv_show in root.findall('tv_show'):
#Function that prepares the values to be added to the graph as a URI or Literal
    show_id = tv_show.attrib["id"]
def prepareValue(row):
    title = tv_show.find("title").text
if row == None: #none type
value = Literal(row)
elif isinstance(row, str) and re.match(r'\d{4}-\d{2}-\d{2}', row): #date
value = Literal(row, datatype=XSD.date)
elif isinstance(row, bool): #boolean value (true / false)
value = Literal(row, datatype=XSD.boolean)
elif isinstance(row, int): #integer
value = Literal(row, datatype=XSD.integer)
elif isinstance(row, str): #string
value = URIRef(ex + row.replace('"', '').replace(" ", "_").replace(",","").replace("-", "_"))
elif isinstance(row, float): #float
value = Literal(row, datatype=XSD.float)


    g.add((URIRef(ex + show_id), ex.title, Literal(title, datatype=XSD.string)))
return value
    g.add((URIRef(ex + show_id), RDF.type, ex.TV_Show))


    for actor in tv_show.findall("actor"):
#Convert the non-semantic CSV dataset into a semantic RDF
        first_name = actor.find("firstname").text
def csv_to_rdf(df):
        last_name = actor.find("lastname").text
for index, row in df.iterrows():
        full_name = first_name + "_" + last_name
id = URIRef(ex + "Investigation_" + str(index))
       
investigation = prepareValue(row["investigation"])
        g.add((URIRef(ex + show_id), ex.stars, URIRef(ex + full_name)))
investigation_start = prepareValue(row["investigation-start"])
        g.add((URIRef(ex + full_name), ex.starsIn, URIRef(title)))
investigation_end = prepareValue(row["investigation-end"])
        g.add((URIRef(ex + full_name), RDF.type, ex.Actor))
investigation_days = prepareValue(row["investigation-days"])
indictment_days = prepareValue(row["indictment-days "])
cp_date = prepareValue(row["cp-date"])
cp_days = prepareValue(row["cp-days"])
overturned = prepareValue(row["overturned"])
pardoned = prepareValue(row["pardoned"])
american = prepareValue(row["american"])
outcome = prepareValue(row["type"])
name_ex = prepareValue(row["name"])
president_ex = prepareValue(row["president"])


print(g.serialize(format="turtle").decode())
#Spotlight Search
</syntaxhighlight>
name = annotate_entity(str(row['name']))
                # Removing the period as some presidents won't be found with it
president = annotate_entity(str(row['president']).replace(".", ""))
#Adds the tripples to the graph
g.add((id, RDF.type, ex.Investigation))
g.add((id, ex.investigation, investigation))
g.add((id, ex.investigation_start, investigation_start))
g.add((id, ex.investigation_end, investigation_end))
g.add((id, ex.investigation_days, investigation_days))
g.add((id, ex.indictment_days, indictment_days))
g.add((id, ex.cp_date, cp_date))
g.add((id, ex.cp_days, cp_days))
g.add((id, ex.overturned, overturned))
g.add((id, ex.pardoned, pardoned))
g.add((id, ex.american, american))
g.add((id, ex.outcome, outcome))


#Spotlight search
#Name
try:
g.add((id, ex.person, URIRef(name[0]["URI"])))
except:
g.add((id, ex.person, name_ex))


==RDFS==
#President
try:
g.add((id, ex.president, URIRef(president[0]["URI"])))
except:
g.add((id, ex.president, president_ex))


===RDFS-plus (OWL) Properties===
csv_to_rdf(df)
<syntaxhighlight>
print(g.serialize())
g.add((ex.married, RDF.type, OWL.SymmetricProperty))
g.add((ex.married, RDF.type, OWL.IrreflexiveProperty))
g.add((ex.livesWith, RDF.type, OWL.ReflexiveProperty))
g.add((ex.livesWith, RDF.type, OWL.SymmetricProperty))
g.add((ex.sibling, RDF.type, OWL.TransitiveProperty))
g.add((ex.sibling, RDF.type, OWL.SymmetricProperty))
g.add((ex.sibling, RDF.type, OWL.IrreflexiveProperty))
g.add((ex.hasFather, RDF.type, OWL.FunctionalProperty))
g.add((ex.hasFather, RDF.type, OWL.AsymmetricProperty))
g.add((ex.hasFather, RDF.type, OWL.IrreflexiveProperty))
g.add((ex.fatherOf, RDF.type, OWL.AsymmetricProperty))
g.add((ex.fatherOf, RDF.type, OWL.IrreflexiveProperty))


# Sometimes there is no definite answer, and it comes down to how we want to model our properties
# e.g is livesWith a transitive property? Usually yes, but we can also want to specify that a child lives with both of her divorced parents.
# which means that: (mother livesWith child % child livesWith father) != mother livesWith father. Which makes it non-transitive.
</syntaxhighlight>
</syntaxhighlight>


===RDFS inference with RDFLib===
==SHACL (Lab 6)==
You can use the OWL-RL package to add inference capabilities to RDFLib. It can be installed using the pip install command:
<syntaxhighlight>
<syntaxhighlight>
pip install owlrl
</syntaxhighlight>
Or download it from [https://github.com/RDFLib/OWL-RL GitHub] and copy the ''owlrl'' subfolder into your project folder next to your Python files.


[https://owl-rl.readthedocs.io/en/latest/owlrl.html OWL-RL documentation.]
from pyshacl import validate
from rdflib import Graph


Example program to get you started. In this example we are creating the graph using sparql.update, but it is also possible to parse the data from a file.
data_graph = Graph()
<syntaxhighlight>
# parses the Turtle examples from the lab
import rdflib.plugins.sparql.update
data_graph.parse("data_graph.ttl")
import owlrl.RDFSClosure


g = rdflib.Graph()
# Remember to test you need to change the rules so they conflict with the data graph (or vice versa). For example, change "exactly one name" to have exactly two, and see the output
shape_graph = """
@prefix ex: <http://example.org/> .
@prefix foaf: <http://xmlns.com/foaf/0.1/> .
@prefix sh: <http://www.w3.org/ns/shacl#> .
@prefix xsd: <http://www.w3.org/2001/XMLSchema#> .
@prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> .


ex = rdflib.Namespace('http://example.org#')
ex:LabTasks_Shape
g.bind('', ex)
    a sh:NodeShape ;
    sh:targetClass ex:PersonUnderInvestigation ;
    sh:property [
        sh:path foaf:name ;
        sh:minCount 1 ; #Every person under investigation has exactly one name.  
        sh:maxCount 1 ; #Every person under investigation has exactly one name.
        sh:datatype rdf:langString ; #All person names must be language-tagged
    ] ;
    sh:property [
        sh:path ex:chargedWith ;
        sh:nodeKind sh:IRI ; #The object of a charged with property must be a URI.
        sh:class ex:Offense ; #The object of a charged with property must be an offense.
    ] .


g.update("""
# --- If you have more time tasks ---
PREFIX ex: <http://example.org#>
ex:MoreTime_Shape rdf:type sh:NodeShape;
PREFIX owl: <http://www.w3.org/2002/07/owl#>
    sh:targetClass ex:Indictment;
INSERT DATA {
   
     ex:Socrates rdf:type ex:Man .
     # The only allowed values for ex:american are true, false or unknown.
     ex:Man rdfs:subClassOf ex:Mortal .
    sh:property [
}""")
        sh:path ex:american;
        sh:pattern "(true|false|unknown)" ;
    ] ;
   
    # The value of a property that counts days must be an integer.
     sh:property [
        sh:path ex:indictment_days;
        sh:datatype xsd:integer;
    ] ; 
    sh:property [
        sh:path ex:investigation_days;
        sh:datatype xsd:integer;
    ] ;
   
    # The value of a property that indicates a start date must be xsd:date.
    sh:property [
        sh:path ex:investigation_start;
        sh:datatype xsd:date;
    ] ;


rdfs = owlrl.RDFSClosure.RDFS_Semantics(g, False, False, False)
    # The value of a property that indicates an end date must be xsd:date or unknown (tip: you can use sh:or (...) ).
# RDF_Semantics parameters:
    sh:property [
# - graph (rdflib.Graph) – The RDF graph to be extended.
        sh:path ex:investigation_end;
# - axioms (bool) – Whether (non-datatype) axiomatic triples should be added or not.
        sh:or (
# - daxioms (bool) – Whether datatype axiomatic triples should be added or not.
        [ sh:datatype xsd:date ]
# - rdfs (bool) – Whether RDFS inference is also done (used in subclassed only).
        [ sh:hasValue "unknown" ]
# For now, you will in most cases use all False in RDFS_Semtantics.
    )] ;
   
    # Every indictment must have exactly one FOAF name for the investigated person.
    sh:property [
        sh:path foaf:name;
        sh:minCount 1;
        sh:maxCount 1;
    ] ;
   
    # Every indictment must have exactly one investigated person property, and that person must have the type ex:PersonUnderInvestigation.
    sh:property [
        sh:path ex:investigatedPerson ;
        sh:minCount 1 ;
        sh:maxCount 1 ;
        sh:class ex:PersonUnderInvestigation ;
        sh:nodeKind sh:IRI ;
    ] ;


# Generates the closure of the graph - generates the new entailed triples, but does not add them to the graph.
    # No URI-s can contain hyphens ('-').
rdfs.closure()
    sh:property [
# Adds the new triples to the graph and empties the RDFS triple-container.
        sh:path ex:outcome ;
rdfs.flush_stored_triples()
        sh:nodeKind sh:IRI ;
        sh:pattern "^[^-]*$" ;
    ] ;


# Ask-query to check whether a new triple has been generated from the entailment.
    # Presidents must be identified with URIs.
b = g.query("""
    sh:property [
PREFIX ex: <http://example.org#>
        sh:path ex:president ;
ASK {
        sh:class ex:President ;
    ex:Socrates rdf:type ex:Mortal .
        sh:nodeKind sh:IRI ;
}
    ] .
""")
"""
print('Result: ' + bool(b))
</syntaxhighlight>
 
===Language tagged RDFS labels===
<syntaxhighlight>
from rdflib import Graph, Namespace, Literal
from rdflib.namespace import RDFS
 
g = Graph()
ex = Namespace("http://example.org/")
 
g.add((ex.France, RDFS.label, Literal("Frankrike", lang="no")))
g.add((ex.France, RDFS.label, Literal("France", lang="en")))
g.add((ex.France, RDFS.label, Literal("Francia", lang="es")))


shacl_graph = Graph()
# parses the contents of a shape_graph made in the tasks
shacl_graph.parse(data=shape_graph)


</syntaxhighlight>
# uses pySHACL's validate method to apply the shape_graph constraints to the data_graph
results = validate(
    data_graph,
    shacl_graph=shacl_graph,
    inference='both'
)


==OWL==
# prints out the validation result
===Basic inference with RDFLib===
boolean_value, results_graph, results_text = results


You can use the OWL-RL package again as for Lecture 5.
# print(boolean_value)
print(results_graph.serialize(format='ttl'))
# print(results_text)


Instead of:
#Write a SPARQL query to print out each distinct sh:resultMessage in the results_graph
<syntaxhighlight>
distinct_messages = """
# The next three lines add inferred triples to g.
PREFIX sh: <http://www.w3.org/ns/shacl#>  
rdfs = owlrl.RDFSClosure.RDFS_Semantics(g, False, False, False)
rdfs.closure()
rdfs.flush_stored_triples()
</syntaxhighlight>
you can write this to get both RDFS and basic RDFS Plus / OWL inference:
<syntaxhighlight>
# The next three lines add inferred triples to g.
owl = owlrl.CombinedClosure.RDFS_OWLRL_Semantics(g, False, False, False)
owl.closure()
owl.flush_stored_triples()
</syntaxhighlight>


Example updates and queries:
SELECT DISTINCT ?message WHERE {
<syntaxhighlight>
    [] sh:result ?errorBlankNode.
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
    ?errorBlankNode sh:resultMessage ?message.  
PREFIX owl: <http://www.w3.org/2002/07/owl#>
PREFIX ex: <http://example.org#>


INSERT DATA {
     # Alternativ and cleaner solution, look at https://www.w3.org/TR/sparql11-query/#pp-language (9.1 Property Path Syntax)
     ex:Socrates ex:hasWife ex:Xanthippe .
     # [] sh:result / sh:resultMessage ?message .
     ex:hasHusband owl:inverseOf ex:hasWife .
}
}
</syntaxhighlight>
"""
messages = results_graph.query(distinct_messages)
for row in messages:
    print(row.message)


<syntaxhighlight>
#each sh:resultMessage in the results_graph once, along with the number of times that message has been repeated in the results
ASK {
count_messages = """
  ex:Xanthippe ex:hasHusband ex:Socrates .
PREFIX sh: <http://www.w3.org/ns/shacl#>  
}
</syntaxhighlight>
 
<syntaxhighlight>
ASK {
  ex:Socrates ^ex:hasHusband ex:Xanthippe .
}
</syntaxhighlight>


<syntaxhighlight>
SELECT ?message (COUNT(?node) AS ?num_messages) WHERE {
INSERT DATA {
     [] sh:result ?errorBlankNode .
     ex:hasWife rdfs:subPropertyOf ex:hasSpouse .
     ?errorBlankNode sh:resultMessage ?message ;
     ex:hasSpouse rdf:type owl:SymmetricProperty .  
                    sh:focusNode ?node .
}
}
</syntaxhighlight>
GROUP BY ?message
ORDER BY DESC(?count) ?message
"""


<syntaxhighlight>
messages = results_graph.query(count_messages)
ASK {
for row in messages:
  ex:Socrates ex:hasSpouse ex:Xanthippe .
    print(f"COUNT: {row.num_messages} | MESSAGE: {row.message}")
}
</syntaxhighlight>


<syntaxhighlight>
ASK {
  ex:Socrates ^ex:hasSpouse ex:Xanthippe .
}
</syntaxhighlight>
</syntaxhighlight>


 
==RDFS (Lab 7)==
 
 
 
 
===XML Data for above example===
<syntaxhighlight>
<syntaxhighlight>
<data>
    <tv_show id="1050">
        <title>The_Sopranos</title>
        <actor>
            <firstname>James</firstname>
            <lastname>Gandolfini</lastname>
        </actor>
    </tv_show>
    <tv_show id="1066">
        <title>Seinfeld</title>
        <actor>
            <firstname>Jerry</firstname>
            <lastname>Seinfeld</lastname>
        </actor>
        <actor>
            <firstname>Julia</firstname>
            <lastname>Louis-dreyfus</lastname>
        </actor>
        <actor>
            <firstname>Jason</firstname>
            <lastname>Alexander</lastname>
        </actor>
    </tv_show>
</data>
</syntaxhighlight>


==Lifting HTML to RDF==
import owlrl
<syntaxhighlight>
from rdflib import Graph, RDF, Namespace, FOAF, RDFS
from bs4 import BeautifulSoup as bs, NavigableString
from rdflib import Graph, URIRef, Namespace
from rdflib.namespace import RDF


g = Graph()
g = Graph()
ex = Namespace("http://example.org/")
ex = Namespace('http://example.org/')
 
g.bind("ex", ex)
g.bind("ex", ex)
g.bind("foaf", FOAF)


html = open("tv_shows.html").read()
NS = {
html = bs(html, features="html.parser")
     'ex': ex,
 
     'rdf': RDF,
shows = html.find_all('li', attrs={'class': 'show'})
     'rdfs': RDFS,
for show in shows:
     'foaf': FOAF,
     title = show.find("h3").text
    actors = show.find('ul', attrs={'class': 'actor_list'})
    for actor in actors:
        if isinstance(actor, NavigableString):
            continue
        else:
            actor = actor.text.replace(" ", "_")
            g.add((URIRef(ex + title), ex.stars, URIRef(ex + actor)))
            g.add((URIRef(ex + actor), RDF.type, ex.Actor))
 
     g.add((URIRef(ex + title), RDF.type, ex.TV_Show))
 
 
print(g.serialize(format="turtle").decode())
</syntaxhighlight>
 
===HTML code for the example above===
<syntaxhighlight>
<!DOCTYPE html>
<html>
<head>
     <meta charset="utf-8">
    <title></title>
</head>
<body>
    <div class="tv_shows">
        <ul>
            <li class="show">
                <h3>The_Sopranos</h3>
                <div class="irrelevant_data"></div>
                <ul class="actor_list">
                    <li>James Gandolfini</li>
                </ul>
            </li>
            <li class="show">
                <h3>Seinfeld</h3>
                <div class="irrelevant_data"></div>
                <ul class="actor_list">
                    <li >Jerry Seinfeld</li>
                    <li>Jason Alexander</li>
                    <li>Julia Louis-Dreyfus</li>
                </ul>
            </li>
        </ul>
    </div>
</body>
</html>
</syntaxhighlight>
 
==Web APIs with JSON==
<syntaxhighlight>
import requests
import json
import pprint
 
# Retrieve JSON data from API service URL. Then load it with the json library as a json object.
url = "http://api.geonames.org/postalCodeLookupJSON?postalcode=46020&#country=ES&username=demo"
data = requests.get(url).content.decode("utf-8")
data = json.loads(data)
pprint.pprint(data)
</syntaxhighlight>
 
 
==JSON-LD==
 
<syntaxhighlight>
import rdflib
 
g = rdflib.Graph()
 
example = """
{
  "@context": {
    "name": "http://xmlns.com/foaf/0.1/name",
     "homepage": {
      "@id": "http://xmlns.com/foaf/0.1/homepage",
      "@type": "@id"
    }
  },
  "@id": "http://me.markus-lanthaler.com/",
  "name": "Markus Lanthaler",
  "homepage": "http://www.markus-lanthaler.com/"
}
"""
 
# json-ld parsing automatically deals with @contexts
g.parse(data=example, format='json-ld')
 
# serialisation does expansion by default
for line in g.serialize(format='json-ld').decode().splitlines():
    print(line)
 
# by supplying a context object, serialisation can do compaction
context = {
    "foaf": "http://xmlns.com/foaf/0.1/"
}
}
for line in g.serialize(format='json-ld', context=context).decode().splitlines():
    print(line)
</syntaxhighlight>


#Write a small function that computes the RDFS closure on your graph.
def flush():
    owlrl.DeductiveClosure(owlrl.RDFS_Semantics).expand(g)


<div class="credits" style="text-align: right; direction: ltr; margin-left: 1em;">''INFO216, UiB, 2017-2020. All code examples are [https://creativecommons.org/choose/zero/ CC0].'' </div>
#Rick Gates was charged with money laundering and tax evasion.
g.add((ex.Rick_Gates, ex.chargedWith, ex.MoneyLaundering))
g.add((ex.Rick_Gates, ex.chargedWith, ex.TaxEvasion))


==OWL - Complex Classes and Restrictions==
#When one thing that is charged with another thing,
<syntaxhighlight>
g.add((ex.chargedWith, RDFS.domain, ex.PersonUnderInvestigation))  #the first thing is a person under investigation and
import owlrl
g.add((ex.chargedWith, RDFS.range, ex.Offense))  #the second thing is an offense.
from rdflib import Graph, Literal, Namespace, BNode
from rdflib.namespace import RDF, OWL, RDFS
from rdflib.collection import Collection


g = Graph()
#Write a SPARQL query that checks the RDF type(s) of Rick Gates and money laundering in your RDF graph.
ex = Namespace("http://example.org/")
print('Is Rick Gates a ex:PersonUnderInvestigation:', g.query('ASK {ex:Rick_Gates rdf:type ex:PersonUnderInvestigation}', initNs=NS).askAnswer)
g.bind("ex", ex)
print('Is Money Laundering a ex:Offense:', g.query('ASK {ex:MoneyLaundering rdf:type ex:Offense}', initNs=NS).askAnswer)
g.bind("owl", OWL)
flush()
print('Is Rick Gates a ex:PersonUnderInvestigation:', g.query('ASK {ex:Rick_Gates rdf:type ex:PersonUnderInvestigation}', initNs=NS).askAnswer)
print('Is Money Laundering a ex:Offense:', g.query('ASK {ex:MoneyLaundering rdf:type ex:Offense}', initNs=NS).askAnswer)


# a Season is either Autumn, Winter, Spring, Summer
#A person under investigation is a FOAF person
seasons = BNode()
g.add((ex.PersonUnderInvestigation, RDFS.subClassOf, FOAF.Person))
Collection(g, seasons, [ex.Winter, ex.Autumn, ex.Spring, ex.Summer])
print('Is Rick Gates a foaf:Person:', g.query('ASK {ex:Rick_Gates rdf:type foaf:Person}', initNs=NS).askAnswer)
g.add((ex.Season, OWL.oneOf, seasons))
flush()
print('Is Rick Gates a foaf:Person:', g.query('ASK {ex:Rick_Gates rdf:type foaf:Person}', initNs=NS).askAnswer)


# A Parent is a Father or Mother
#Paul Manafort was convicted for tax evasion.
b = BNode()
g.add((ex.Paul_Manafort, ex.convictedFor, ex.TaxEvasion))
Collection(g, b, [ex.Father, ex.Mother])
#the first thing is also charged with the second thing
g.add((ex.Parent, OWL.unionOf, b))
g.add((ex.convictedFor, RDFS.subPropertyOf, ex.chargedWith))
print('Is Paul Manafort charged with Tax Evasion:', g.query('ASK {ex:Paul_Manafort ex:chargedWith ex:TaxEvasion}', initNs=NS).askAnswer)
flush()
print('Is Paul Manafort charged with Tax Evasion:', g.query('ASK {ex:Paul_Manafort ex:chargedWith ex:TaxEvasion}', initNs=NS).askAnswer)


# A Woman is a person who has the "female" gender
print(g.serialize())
br = BNode()
g.add((br, RDF.type, OWL.Restriction))
g.add((br, OWL.onProperty, ex.gender))
g.add((br, OWL.hasValue, ex.Female))
bi = BNode()
Collection(g, bi, [ex.Person, br])
g.add((ex.Woman, OWL.intersectionOf, bi))
 
# A vegetarian is a Person who only eats vegetarian food
br = BNode()
g.add((br, RDF.type, OWL.Restriction))
g.add((br, OWL.onProperty, ex.eats))
g.add((br, OWL.allValuesFrom, ex.VeganFood))
bi = BNode()
Collection(g, bi, [ex.Person, br])
g.add((ex.Vegetarian, OWL.intersectionOf, bi))
 
# A vegetarian is a Person who can not eat meat.
br = BNode()
g.add((br, RDF.type, OWL.Restriction))
g.add((br, OWL.onProperty, ex.eats))
g.add((br, OWL.QualifiedCardinality, Literal(0)))
g.add((br, OWL.onClass, ex.Meat))
bi = BNode()
Collection(g, bi, [ex.Person, br])
g.add((ex.Vegetarian, OWL.intersectionOf, bi))
 
# A Worried Parent is a parent who has at least one sick child
br = BNode()
g.add((br, RDF.type, OWL.Restriction))
g.add((br, OWL.onProperty, ex.hasChild))
g.add((br, OWL.QualifiedMinCardinality, Literal(1)))
g.add((br, OWL.onClass, ex.Sick))
bi = BNode()
Collection(g, bi, [ex.Parent, br])
g.add((ex.WorriedParent, OWL.intersectionOf, bi))
 
# using the restriction above, If we now write...:
g.add((ex.Bob, RDF.type, ex.Parent))
g.add((ex.Bob, ex.hasChild, ex.John))
g.add((ex.John, RDF.type, ex.Sick))
# ...we can infer with owl reasoning that Bob is a worried Parent even though we didn't specify it ourselves because Bob fullfills the restriction and Parent requirements.


</syntaxhighlight>
</syntaxhighlight>
==Protege-OWL reasoning with HermiT==
[[:File:DL-reasoning-RoyalFamily-final.owl.txt | Example file]] from Lecture 13 about OWL-DL, rules and reasoning.
-->

Revision as of 14:00, 21 March 2023

This page will be updated with Python examples related to the labs as the course progresses.

Examples from the lectures

Lecture 1: Introduction to KGs

Turtle example:

@prefix ex: <http://example.org/> .
ex:Roger_Stone
    ex:name "Roger Stone" ;
    ex:occupation ex:lobbyist ;
    ex:significant_person ex:Donald_Trump .
ex:Donald_Trump
    ex:name "Donald Trump" .

Lecture 2: RDF

Blank nodes for anonymity, or when we have not decided on a URI:

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

EX = Namespace('http://example.org/')

g = Graph()
g.bind('ex', EX)  # this is why the line '@prefix ex: <http://example.org/> .'
                  # and the 'ex.' prefix are used when we print out Turtle later

robertMueller = BNode()
g.add((robertMueller, RDF.type, EX.Human))
g.add((robertMueller, FOAF.name, Literal('Robert Mueller', lang='en')))
g.add((robertMueller, EX.position_held, Literal('Director of the Federal Bureau of Investigation', lang='en')))

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

Blank nodes used to group related properties:

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

EX = Namespace('http://example.org/')

g = Graph()
g.bind('ex', EX)

# This is a task in Exercise 2

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

Literals:

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

EX = Namespace('http://example.org/')

g = Graph()
g.bind('ex', EX)

g.add((EX.Robert_Mueller, RDF.type, EX.Human))
g.add((EX.Robert_Mueller, FOAF.name, Literal('Robert Mueller', lang='en')))
g.add((EX.Robert_Mueller, FOAF.name, Literal('رابرت مولر', lang='fa')))
g.add((EX.Robert_Mueller, DC.description, Literal('sixth director of the FBI', datatype=XSD.string)))
g.add((EX.Robert_Mueller, EX.start_time, Literal(2001, datatype=XSD.integer)))

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

Alternative container (open):

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

EX = Namespace('http://example.org/')

g = Graph()
g.bind('ex', EX)

muellerReportArchives = BNode()
g.add((muellerReportArchives, RDF.type, RDF.Alt))

archive1 = 'https://archive.org/details/MuellerReportVolume1Searchable/' \
                    'Mueller%20Report%20Volume%201%20Searchable/'
archive2 = 'https://edition.cnn.com/2019/04/18/politics/full-mueller-report-pdf/index.html'
archive3 = 'https://www.politico.com/story/2019/04/18/mueller-report-pdf-download-text-file-1280891'

g.add((muellerReportArchives, RDFS.member, Literal(archive1, datatype=XSD.anyURI)))
g.add((muellerReportArchives, RDFS.member, Literal(archive2, datatype=XSD.anyURI)))
g.add((muellerReportArchives, RDFS.member, Literal(archive3, datatype=XSD.anyURI)))

g.add((EX.Mueller_Report, RDF.type, FOAF.Document))
g.add((EX.Mueller_Report, DC.contributor, EX.Robert_Mueller))
g.add((EX.Mueller_Report, SCHEMA.archivedAt, muellerReportArchives))

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

Sequence container (open):

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

EX = Namespace('http://example.org/')

g = Graph()
g.bind('ex', EX)

donaldTrumpSpouses = BNode()
g.add((donaldTrumpSpouses, RDF.type, RDF.Seq))
g.add((donaldTrumpSpouses, RDF._1, EX.IvanaTrump))
g.add((donaldTrumpSpouses, RDF._2, EX.MarlaMaples))
g.add((donaldTrumpSpouses, RDF._3, EX.MelaniaTrump))

g.add((EX.Donald_Trump, SCHEMA.spouse, donaldTrumpSpouses))

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

Collection (closed list):

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

EX = Namespace('http://example.org/')

g = Graph()
g.bind('ex', EX)

from rdflib.collection import Collection

g = Graph()
g.bind('ex', EX)

donaldTrumpSpouses = BNode()
Collection(g, donaldTrumpSpouses, [
    EX.IvanaTrump, EX.MarlaMaples, EX.MelaniaTrump
])
g.add((EX.Donald_Trump, SCHEMA.spouse, donaldTrumpSpouses))

print(g.serialize(format='turtle'))
g.serialize(destination='s02_Donald_Trump_spouses_list.ttl', format='turtle')

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

Example lab solutions

Getting started (Lab 1)

from rdflib import Graph, Namespace

g = Graph()

ex = Namespace('http://example.org/')

g.bind("ex", ex)

#The Mueller Investigation was lead by Robert Mueller.
g.add((ex.Mueller_Investigation, ex.leadBy, ex.Robert_Muller))

#It involved Paul Manafort, Rick Gates, George Papadopoulos, Michael Flynn, and Roger Stone.
g.add((ex.Mueller_Investigation, ex.involved, ex.Paul_Manafort))
g.add((ex.Mueller_Investigation, ex.involved, ex.Rick_Gates))
g.add((ex.Mueller_Investigation, ex.involved, ex.George_Papadopoulos))
g.add((ex.Mueller_Investigation, ex.involved, ex.Michael_Flynn))
g.add((ex.Mueller_Investigation, ex.involved, ex.Michael_Cohen))
g.add((ex.Mueller_Investigation, ex.involved, ex.Roger_Stone))

# --- Paul Manafort ---
#Paul Manafort was business partner of Rick Gates.
g.add((ex.Paul_Manafort, ex.businessManager, ex.Rick_Gates))
# He was campaign chairman for Trump
g.add((ex.Paul_Manafort, ex.campaignChairman, ex.Donald_Trump))

# He was charged with money laundering, tax evasion, and foreign lobbying.
g.add((ex.Paul_Manafort, ex.chargedWith, ex.MoneyLaundering))
g.add((ex.Paul_Manafort, ex.chargedWith, ex.TaxEvasion))
g.add((ex.Paul_Manafort, ex.chargedWith, ex.ForeignLobbying))

# He was convicted for bank and tax fraud.
g.add((ex.Paul_Manafort, ex.convictedFor, ex.BankFraud))
g.add((ex.Paul_Manafort, ex.convictedFor, ex.TaxFraud))

# He pleaded guilty to conspiracy.
g.add((ex.Paul_Manafort, ex.pleadGuiltyTo, ex.Conspiracy))
# He was sentenced to prison.
g.add((ex.Paul_Manafort, ex.sentencedTo, ex.Prison))
# He negotiated a plea agreement.
g.add((ex.Paul_Manafort, ex.negoiated, ex.PleaBargain))

# --- Rick Gates ---
#Rick Gates was charged with money laundering, tax evasion and foreign lobbying.
g.add((ex.Rick_Gates, ex.chargedWith, ex.MoneyLaundering))
g.add((ex.Rick_Gates, ex.chargedWith, ex.TaxEvasion))
g.add((ex.Rick_Gates, ex.chargedWith, ex.ForeignLobbying))

#He pleaded guilty to conspiracy and lying to FBI.
g.add((ex.Rick_Gates, ex.pleadGuiltyTo, ex.Conspiracy))
g.add((ex.Rick_Gates, ex.pleadGuiltyTo, ex.LyingToFBI))

#Use the serialize method to write out the model in different formats on screen
print(g.serialize(format="ttl"))
# g.serialize("lab1.ttl", format="ttl") #or to file

#Loop through the triples in the model to print out all triples that have pleading guilty as predicate
for subject, object in g[ : ex.pleadGuiltyTo : ]:
    print(subject, ex.pleadGuiltyTo, object)

# Michael Cohen, Michael Flynn and the lying is part of lab 2 and therefore the answer is not provided this week 

#Write a method (function) that submits your model for rendering and saves the returned image to file.
import requests
import shutil

def graphToImage(graph):
    data = {"rdf":graph, "from":"ttl", "to":"png"}
    link = "http://www.ldf.fi/service/rdf-grapher"
    response = requests.get(link, params = data, stream=True)
    # print(response.content)
    print(response.raw)
    with open("lab1.png", "wb") as fil:
        shutil.copyfileobj(response.raw, fil)

graph = g.serialize(format="ttl")
graphToImage(graph)

RDF programming with RDFlib (Lab 2)

from rdflib import Graph, URIRef, Namespace, Literal, XSD, BNode
from rdflib.collection import Collection

g = Graph()
g.parse("lab1.ttl", format="ttl") #Retrives the triples from lab 1

ex = Namespace('http://example.org/')

# --- Michael Cohen ---
#Michael Cohen was Donald Trump's attorney.
g.add((ex.Michael_Cohen, ex.attorneyTo, ex.Donald_Trump))
#He pleaded guilty to lying to the FBI.
g.add((ex.Michael_Cohen, ex.pleadGuiltyTo, ex.LyingToCongress))

# --- Michael Flynn ---
#Michael Flynn was adviser to Trump.
g.add((ex.Michael_Flynn, ex.adviserTo, ex.Donald_Trump))
#He pleaded guilty to lying to the FBI.
g.add((ex.Michael_Flynn, ex.pleadGuiltyTo, ex.LyingToFBI))
# He negotiated a plea agreement.
g.add((ex.Michael_Flynn, ex.negoiated, ex.PleaBargain)) 

#How can you modify your knowledge graph to account for the different lying?
#Remove these to not have duplicates
g.remove((ex.Michael_Flynn, ex.pleadGuiltyTo, ex.LyingToFBI)) 
g.remove((ex.Michael_Flynn, ex.negoiated, ex.PleaBargain))
g.remove((ex.Rick_Gates, ex.pleadGuiltyTo, ex.LyingToFBI))
g.remove((ex.Rick_Gates, ex.pleadGuiltyTo, ex.Conspiracy))
g.remove((ex.Rick_Gates, ex.chargedWith, ex.ForeignLobbying))
g.remove((ex.Rick_Gates, ex.chargedWith, ex.MoneyLaundering))
g.remove((ex.Rick_Gates, ex.chargedWith, ex.TaxEvasion))
g.remove((ex.Michael_Cohen, ex.pleadGuiltyTo, ex.LyingToCongress))

# --- Michael Flynn ---
FlynnLying = BNode() 
g.add((FlynnLying, ex.crime, ex.LyingToFBI))
g.add((FlynnLying, ex.pleadGulityOn, Literal("2017-12-1", datatype=XSD.date)))
g.add((FlynnLying, ex.liedAbout, Literal("His communications with a former Russian ambassador during the presidential transition", datatype=XSD.string)))
g.add((FlynnLying, ex.pleaBargain, Literal("true", datatype=XSD.boolean)))
g.add((ex.Michael_Flynn, ex.pleadGuiltyTo, FlynnLying))

# --- Rick Gates ---
GatesLying = BNode()
Crimes = BNode()
Charged = BNode()
Collection(g, Crimes, [ex.LyingToFBI, ex.Conspiracy])
Collection(g, Charged, [ex.ForeignLobbying, ex.MoneyLaundering, ex.TaxEvasion])
g.add((GatesLying, ex.crime, Crimes))
g.add((GatesLying, ex.chargedWith, Charged))
g.add((GatesLying, ex.pleadGulityOn, Literal("2018-02-23", datatype=XSD.date)))
g.add((GatesLying, ex.pleaBargain, Literal("true", datatype=XSD.boolean)))
g.add((ex.Rick_Gates, ex.pleadGuiltyTo, GatesLying))

# --- Michael Cohen ---
CohenLying = BNode()
g.add((CohenLying, ex.crime, ex.LyingToCongress))
g.add((CohenLying, ex.liedAbout, ex.TrumpRealEstateDeal))
g.add((CohenLying, ex.prosecutorsAlleged, Literal("In an August 2017 letter Cohen sent to congressional committees investigating Russian election interference, he falsely stated that the project ended in January 2016", datatype=XSD.string)))
g.add((CohenLying, ex.mullerInvestigationAlleged, Literal("Cohen falsely stated that he had never agreed to travel to Russia for the real estate deal and that he did not recall any contact with the Russian government about the project", datatype=XSD.string)))
g.add((CohenLying, ex.pleadGulityOn, Literal("2018-11-29", datatype=XSD.date)))
g.add((CohenLying, ex.pleaBargain, Literal("true", datatype=XSD.boolean)))
g.add((ex.Michael_Cohen, ex.pleadGuiltyTo, CohenLying))

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

#Save (serialize) your graph to a Turtle file.
# g.serialize("lab2.ttl", format="ttl")

#Add a few triples to the Turtle file with more information about Donald Trump.
'''
ex:Donald_Trump ex:address [ ex:city ex:Palm_Beach ;
            ex:country ex:United_States ;
            ex:postalCode 33480 ;
            ex:residence ex:Mar_a_Lago ;
            ex:state ex:Florida ;
            ex:streetName "1100 S Ocean Blvd"^^xsd:string ] ;
    ex:previousAddress [ ex:city ex:Washington_DC ;
            ex:country ex:United_States ;
            ex:phoneNumber "1 202 456 1414"^^xsd:integer ;
            ex:postalCode "20500"^^xsd:integer ;
            ex:residence ex:The_White_House ;
            ex:streetName "1600 Pennsylvania Ave."^^xsd:string ];
    ex:marriedTo ex:Melania_Trump;
    ex:fatherTo (ex:Ivanka_Trump ex:Donald_Trump_Jr ex: ex:Tiffany_Trump ex:Eric_Trump ex:Barron_Trump).
'''

#Read (parse) the Turtle file back into a Python program, and check that the new triples are there
def serialize_Graph():
    newGraph = Graph()
    newGraph.parse("lab2.ttl")
    print(newGraph.serialize())

# serialize_Graph() #Don't need this to run until after adding the triples above to the ttl file

#Write a method (function) that starts with Donald Trump prints out a graph depth-first to show how the other graph nodes are connected to him
visited_nodes = set()

def create_Tree(model, nodes):
    #Traverse the model breadth-first to create the tree.
    global visited_nodes
    tree = Graph()
    children = set()
    visited_nodes |= set(nodes)
    for s, p, o in model:
        if s in nodes and o not in visited_nodes:
            tree.add((s, p, o))
            visited_nodes.add(o)
            children.add(o)
        if o in nodes and s not in visited_nodes:
            invp = URIRef(f'{p}_inv') #_inv represents inverse of
            tree.add((o, invp, s))
            visited_nodes.add(s)
            children.add(s)
    if len(children) > 0:
        children_tree = create_Tree(model, children)
        for triple in children_tree:
            tree.add(triple)
    return tree

def print_Tree(tree, root, indent=0):
    #Print the tree depth-first.
    print(str(root))
    for s, p, o in tree:
        if s==root:
            print('    '*indent + '  ' + str(p), end=' ')
            print_Tree(tree, o, indent+1)
    
tree = create_Tree(g, [ex.Donald_Trump])
print_Tree(tree, ex.Donald_Trump)

SPARQL Programming (Lab 4)

NOTE: These tasks were performed on the old dataset, with the new dataset, some of these answers would be different.

from rdflib import Graph, Namespace, RDF, FOAF
from SPARQLWrapper import SPARQLWrapper, JSON, POST, GET, TURTLE

g = Graph()
g.parse("Russia_investigation_kg.ttl")

# ----- RDFLIB -----
ex = Namespace('http://example.org#')

NS = {
    '': ex,
    'rdf': RDF,
    'foaf': FOAF,
}

# Print out a list of all the predicates used in your graph.
task1 = g.query("""
SELECT DISTINCT ?p WHERE{
    ?s ?p ?o .
}
""", initNs=NS)

print(list(task1))

# Print out a sorted list of all the presidents represented in your graph.
task2 = g.query("""
SELECT DISTINCT ?president WHERE{
    ?s :president ?president .
}
ORDER BY ?president
""", initNs=NS)

print(list(task2))

# Create dictionary (Python dict) with all the represented presidents as keys. For each key, the value is a list of names of people indicted under that president.
task3_dic = {}

task3 = g.query("""
SELECT ?president ?person WHERE{
    ?s :president ?president;
       :name ?person;
       :outcome :indictment.
}
""", initNs=NS)

for president, person in task3:
    if president not in task3_dic:
        task3_dic[president] = [person]
    else:
        task3_dic[president].append(person)

print(task3_dic)

# Use an ASK query to investigate whether Donald Trump has pardoned more than 5 people.

# This task is a lot trickier than it needs to be. As far as I'm aware RDFLib has no HAVING support, so a query like this:
task4 = g.query("""
ASK {
  	SELECT (COUNT(?s) as ?count) WHERE{
    	?s :pardoned :true;
   	   :president :Bill_Clinton  .
    }
    HAVING (?count > 5)
}
""", initNs=NS)

print(task4.askAnswer)

# Which works fine in Blazegraph and is a valid SPARQL query will always provide false in RDFLib, cause it uses HAVING. Instead you have to use a nested SELECT query like below, where you use FILTER instead of HAVING. Donald Trump has no pardons, so I have instead chosen Bill Clinton (which has 13 pardons) to check if the query works. 

task4 = g.query("""
    ASK{
        SELECT ?count WHERE{{
  	        SELECT (COUNT(?s) as ?count) WHERE{
    	        ?s :pardoned :true;
                   :president :Bill_Clinton  .
                }}
        FILTER (?count > 5) 
        }
    }
""", initNs=NS)

print(task4.askAnswer)

# Use a DESCRIBE query to create a new graph with information about Donald Trump. Print out the graph in Turtle format.

# By all accounts, it seems DESCRIBE queries are yet to be implemented in RDFLib, but they are attempting to implement it: https://github.com/RDFLib/rdflib/pull/2221 (Issue and proposed solution raised) & https://github.com/RDFLib/rdflib/commit/2325b4a81724c1ccee3a131067db4fbf9b4e2629 (Solution committed to RDFLib). This solution does not work. However, this proposed solution should work if DESCRIBE is implemented in RDFLib

# task5 = g.query(""" 
# DESCRIBE :Donald_Trump
# """, initNs=NS)

# print(task5.serialize())

# ----- SPARQLWrapper -----

namespace = "kb" #Default namespace
sparql = SPARQLWrapper("http://localhost:9999/blazegraph/namespace/"+ namespace + "/sparql") #Replace localhost:9999 with your URL

# The current dates are URIs, we would want to change them to Literals with datatype "date" for task 1 & 2
update_str = """
    PREFIX ns1: <http://example.org#>

    DELETE {
        ?s ns1:cp_date ?cp;
            ns1:investigation_end ?end;
            ns1:investigation_start ?start.
    }
    INSERT{
        ?s ns1:cp_date ?cpDate;
            ns1:investigation_end ?endDate;
            ns1:investigation_start ?startDate.
    }
    WHERE{
        ?s ns1:cp_date ?cp . #Date conviction was recieved
        BIND (replace(str(?cp), str(ns1:), "")  AS ?cpRemoved)
        BIND (STRDT(STR(?cpRemoved), xsd:date) AS ?cpDate)
        
        ?s ns1:investigation_end ?end . #Investigation End
        BIND (replace(str(?end), str(ns1:), "")  AS ?endRemoved)
        BIND (STRDT(STR(?endRemoved), xsd:date) AS ?endDate)
        
        ?s ns1:investigation_start ?start . #Investigation Start
        BIND (replace(str(?start), str(ns1:), "")  AS ?startRemoved)
        BIND (STRDT(STR(?startRemoved), xsd:date) AS ?startDate)
}"""

sparql.setQuery(update_str)
sparql.setMethod(POST)
sparql.query()

# Ask whether there was an ongoing indictment on the date 1990-01-01.
sparql.setQuery("""
    PREFIX ns1: <http://example.org#>
    ASK {
        SELECT ?end ?start
        WHERE{
            ?s ns1:investigation_end ?end;
               ns1:investigation_start ?start;
               ns1:outcome ns1:indictment.
            FILTER(?start <= "1990-01-01"^^xsd:date && ?end >= "1990-01-01"^^xsd:date) 
	    }
    }
""")
sparql.setReturnFormat(JSON)
results = sparql.query().convert()
print(f"Are there any investigation on the 1990-01-01: {results['boolean']}")

# List ongoing indictments on that date 1990-01-01.
sparql.setQuery("""
    PREFIX ns1: <http://example.org#>
    SELECT ?s
    WHERE{
        ?s ns1:investigation_end ?end;
           ns1:investigation_start ?start;
           ns1:outcome ns1:indictment.
        FILTER(?start <= "1990-01-01"^^xsd:date && ?end >= "1990-01-01"^^xsd:date) 
    }
""")

sparql.setReturnFormat(JSON)
results = sparql.query().convert()

print("The ongoing investigations on the 1990-01-01 are:")
for result in results["results"]["bindings"]:
    print(result["s"]["value"])

# Describe investigation number 100 (muellerkg:investigation_100).
sparql.setQuery("""
    PREFIX ns1: <http://example.org#>
    DESCRIBE ns1:investigation_100
""")

sparql.setReturnFormat(TURTLE)
results = sparql.query().convert()

print(results.serialize())

# Print out a list of all the types used in your graph.
sparql.setQuery("""
    PREFIX ns1: <http://example.org#>
    PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>

    SELECT DISTINCT ?types
    WHERE{
        ?s rdf:type ?types . 
    }
""")

sparql.setReturnFormat(JSON)
results = sparql.query().convert()

rdf_Types = []

for result in results["results"]["bindings"]:
    rdf_Types.append(result["types"]["value"])

print(rdf_Types)

# Update the graph to that every resource that is an object in a muellerkg:investigation triple has the rdf:type muellerkg:Investigation.
update_str = """
    PREFIX ns1: <http://example.org#>
    PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>

    INSERT{
        ?invest rdf:type ns1:Investigation .
    }
    WHERE{
        ?s ns1:investigation ?invest .
}"""

sparql.setQuery(update_str)
sparql.setMethod(POST)
sparql.query()

#To Test
sparql.setQuery("""
    prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
    PREFIX ns1: <http://example.org#>

    ASK{
        ns1:watergate rdf:type ns1:Investigation.
    }
""")

sparql.setReturnFormat(JSON)
results = sparql.query().convert()
print(results['boolean'])

# Update the graph to that every resource that is an object in a muellerkg:person triple has the rdf:type muellerkg:IndictedPerson.
update_str = """
    PREFIX ns1: <http://example.org#>
    PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>

    INSERT{
        ?person rdf:type ns1:IndictedPerson .
    }
    WHERE{
        ?s ns1:person ?person .
}"""

sparql.setQuery(update_str)
sparql.setMethod(POST)
sparql.query()

#To test, run the query in the above task, replacing the ask query with e.g. ns1:Deborah_Gore_Dean rdf:type ns1:IndictedPerson

# Update the graph so all the investigation nodes (such as muellerkg:watergate) become the subject in a dc:title triple with the corresponding string (watergate) as the literal.
update_str = """
    PREFIX ns1: <http://example.org#>
    PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
    PREFIX dc: <http://purl.org/dc/elements/1.1/>

    INSERT{
        ?invest dc:title ?investString.
    }
    WHERE{
        ?s ns1:investigation ?invest .
        BIND (replace(str(?invest), str(ns1:), "")  AS ?investString)
}"""

sparql.setQuery(update_str)
sparql.setMethod(POST)
sparql.query()

#Same test as above, replace it with e.g. ns1:watergate dc:title "watergate"

# Print out a sorted list of all the indicted persons represented in your graph.
sparql.setQuery("""
    PREFIX ns1: <http://example.org#>
    PREFIX foaf: <http://xmlns.com/foaf/0.1/>

    SELECT ?name
    WHERE{
    ?s  ns1:person ?name;
        ns1:outcome ns1:indictment.
    }
    ORDER BY ?name
""")

sparql.setReturnFormat(JSON)
results = sparql.query().convert()

names = []

for result in results["results"]["bindings"]:
    names.append(result["name"]["value"])

print(names)

# Print out the minimum, average and maximum indictment days for all the indictments in the graph.
sparql.setQuery("""
    prefix xsd: <http://www.w3.org/2001/XMLSchema#>
    PREFIX ns1: <http://example.org#>

    SELECT (AVG(?daysRemoved) as ?avg) (MAX(?daysRemoved) as ?max) (MIN(?daysRemoved) as ?min)  WHERE{
        ?s  ns1:indictment_days ?days;
            ns1:outcome ns1:indictment.
    
    BIND (replace(str(?days), str(ns1:), "")  AS ?daysR)
    BIND (STRDT(STR(?daysR), xsd:float) AS ?daysRemoved)
}
""")

sparql.setReturnFormat(JSON)
results = sparql.query().convert()

for result in results["results"]["bindings"]:
    print(f'The longest an investigation lasted was: {result["max"]["value"]}')
    print(f'The shortest an investigation lasted was: {result["min"]["value"]}')
    print(f'The average investigation lasted: {result["avg"]["value"]}')

# Print out the minimum, average and maximum indictment days for all the indictments in the graph per investigation.
sparql.setQuery("""
    prefix xsd: <http://www.w3.org/2001/XMLSchema#>
    PREFIX ns1: <http://example.org#>

    SELECT ?investigation (AVG(?daysRemoved) as ?avg) (MAX(?daysRemoved) as ?max) (MIN(?daysRemoved) as ?min)  WHERE{
    ?s  ns1:indictment_days ?days;
        ns1:outcome ns1:indictment;
        ns1:investigation ?investigation.
    
    BIND (replace(str(?days), str(ns1:), "")  AS ?daysR)
    BIND (STRDT(STR(?daysR), xsd:float) AS ?daysRemoved)
    }
    GROUP BY ?investigation
""")

sparql.setReturnFormat(JSON)
results = sparql.query().convert()

for result in results["results"]["bindings"]:
    print(f'{result["investigation"]["value"]} - min: {result["min"]["value"]}, max: {result["max"]["value"]}, avg: {result["avg"]["value"]}')

CSV To RDF (Lab 5)

#Imports
import re
from pandas import *
from numpy import nan
from rdflib import Graph, Namespace, URIRef, Literal, RDF, XSD, FOAF
from spotlight import SpotlightException, annotate

SERVER = "https://api.dbpedia-spotlight.org/en/annotate"
# Test around with the confidence, and see how many names changes depending on the confidence. However, be aware that anything lower than this (0.83) it will replace James W. McCord and other names that includes James with LeBron James
CONFIDENCE = 0.83 

def annotate_entity(entity, filters={'types': 'DBpedia:Person'}):
	annotations = []
	try:
		annotations = annotate(address=SERVER, text=entity, confidence=CONFIDENCE, filters=filters)
	except SpotlightException as e:
		print(e)
	return annotations

g = Graph()
ex = Namespace("http://example.org/")
g.bind("ex", ex)

#Pandas' read_csv function to load russia-investigation.csv
df = read_csv("russia-investigation.csv")
#Replaces all instances of nan to None type with numpy's nan
df = df.replace(nan, None)

#Function that prepares the values to be added to the graph as a URI or Literal
def prepareValue(row):
	if row == None: #none type
		value = Literal(row)
	elif isinstance(row, str) and re.match(r'\d{4}-\d{2}-\d{2}', row): #date
		value = Literal(row, datatype=XSD.date)
	elif isinstance(row, bool): #boolean value (true / false)
		value = Literal(row, datatype=XSD.boolean)
	elif isinstance(row, int): #integer
		value = Literal(row, datatype=XSD.integer)
	elif isinstance(row, str): #string
		value = URIRef(ex + row.replace('"', '').replace(" ", "_").replace(",","").replace("-", "_"))
	elif isinstance(row, float): #float
		value = Literal(row, datatype=XSD.float)

	return value

#Convert the non-semantic CSV dataset into a semantic RDF 
def csv_to_rdf(df):
	for index, row in df.iterrows():
		id = URIRef(ex + "Investigation_" + str(index))
		investigation = prepareValue(row["investigation"])
		investigation_start = prepareValue(row["investigation-start"])
		investigation_end = prepareValue(row["investigation-end"])
		investigation_days = prepareValue(row["investigation-days"])
		indictment_days = prepareValue(row["indictment-days "])
		cp_date = prepareValue(row["cp-date"])
		cp_days = prepareValue(row["cp-days"])
		overturned = prepareValue(row["overturned"])
		pardoned = prepareValue(row["pardoned"])
		american = prepareValue(row["american"])
		outcome = prepareValue(row["type"])
		name_ex = prepareValue(row["name"])
		president_ex = prepareValue(row["president"])

		#Spotlight Search
		name = annotate_entity(str(row['name']))
                # Removing the period as some presidents won't be found with it
		president = annotate_entity(str(row['president']).replace(".", ""))
		
		#Adds the tripples to the graph
		g.add((id, RDF.type, ex.Investigation))
		g.add((id, ex.investigation, investigation))
		g.add((id, ex.investigation_start, investigation_start))
		g.add((id, ex.investigation_end, investigation_end))
		g.add((id, ex.investigation_days, investigation_days))
		g.add((id, ex.indictment_days, indictment_days))
		g.add((id, ex.cp_date, cp_date))
		g.add((id, ex.cp_days, cp_days))
		g.add((id, ex.overturned, overturned))
		g.add((id, ex.pardoned, pardoned))
		g.add((id, ex.american, american))
		g.add((id, ex.outcome, outcome))

		#Spotlight search
		#Name
		try:
			g.add((id, ex.person, URIRef(name[0]["URI"])))
		except:
			g.add((id, ex.person, name_ex))

		#President
		try:
			g.add((id, ex.president, URIRef(president[0]["URI"])))
		except:
			g.add((id, ex.president, president_ex))

csv_to_rdf(df)
print(g.serialize())

SHACL (Lab 6)

from pyshacl import validate
from rdflib import Graph

data_graph = Graph()
# parses the Turtle examples from the lab
data_graph.parse("data_graph.ttl")

# Remember to test you need to change the rules so they conflict with the data graph (or vice versa). For example, change "exactly one name" to have exactly two, and see the output 
shape_graph = """
@prefix ex: <http://example.org/> .
@prefix foaf: <http://xmlns.com/foaf/0.1/> .
@prefix sh: <http://www.w3.org/ns/shacl#> .
@prefix xsd: <http://www.w3.org/2001/XMLSchema#> .
@prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> .

ex:LabTasks_Shape
    a sh:NodeShape ;
    sh:targetClass ex:PersonUnderInvestigation ;
    sh:property [
        sh:path foaf:name ;
        sh:minCount 1 ; #Every person under investigation has exactly one name. 
        sh:maxCount 1 ; #Every person under investigation has exactly one name.
        sh:datatype rdf:langString ; #All person names must be language-tagged
    ] ;
    sh:property [
        sh:path ex:chargedWith ;
        sh:nodeKind sh:IRI ; #The object of a charged with property must be a URI.
        sh:class ex:Offense ; #The object of a charged with property must be an offense.
    ] .

# --- If you have more time tasks ---
ex:MoreTime_Shape rdf:type sh:NodeShape;
    sh:targetClass ex:Indictment;
    
    # The only allowed values for ex:american are true, false or unknown.
    sh:property [
        sh:path ex:american;
        sh:pattern "(true|false|unknown)" ;
    ] ;
    
    # The value of a property that counts days must be an integer.
    sh:property [
        sh:path ex:indictment_days;
        sh:datatype xsd:integer;
    ] ;   
    sh:property [
        sh:path ex:investigation_days;
        sh:datatype xsd:integer;
    ] ;
    
    # The value of a property that indicates a start date must be xsd:date.
    sh:property [
        sh:path ex:investigation_start;
        sh:datatype xsd:date;
    ] ;

    # The value of a property that indicates an end date must be xsd:date or unknown (tip: you can use sh:or (...) ).
    sh:property [
        sh:path ex:investigation_end;
        sh:or (
         [ sh:datatype xsd:date ]
         [ sh:hasValue "unknown" ]
    )] ;
    
    # Every indictment must have exactly one FOAF name for the investigated person.
    sh:property [
        sh:path foaf:name;
        sh:minCount 1;
        sh:maxCount 1;
    ] ;
    
    # Every indictment must have exactly one investigated person property, and that person must have the type ex:PersonUnderInvestigation.
    sh:property [
        sh:path ex:investigatedPerson ;
        sh:minCount 1 ;
        sh:maxCount 1 ;
        sh:class ex:PersonUnderInvestigation ;
        sh:nodeKind sh:IRI ;
    ] ;

    # No URI-s can contain hyphens ('-').
    sh:property [
        sh:path ex:outcome ;
        sh:nodeKind sh:IRI ;
        sh:pattern "^[^-]*$" ;
    ] ;

    # Presidents must be identified with URIs.
    sh:property [
        sh:path ex:president ;
        sh:class ex:President ;
        sh:nodeKind sh:IRI ;
    ] .
"""

shacl_graph = Graph()
# parses the contents of a shape_graph made in the tasks
shacl_graph.parse(data=shape_graph)

# uses pySHACL's validate method to apply the shape_graph constraints to the data_graph
results = validate(
    data_graph,
    shacl_graph=shacl_graph,
    inference='both'
)

# prints out the validation result
boolean_value, results_graph, results_text = results

# print(boolean_value)
print(results_graph.serialize(format='ttl'))
# print(results_text)

#Write a SPARQL query to print out each distinct sh:resultMessage in the results_graph
distinct_messages = """
PREFIX sh: <http://www.w3.org/ns/shacl#> 

SELECT DISTINCT ?message WHERE {
    [] sh:result ?errorBlankNode.
    ?errorBlankNode sh:resultMessage ?message.    

    # Alternativ and cleaner solution, look at https://www.w3.org/TR/sparql11-query/#pp-language (9.1 Property Path Syntax)
    # [] sh:result / sh:resultMessage ?message .
}
"""
messages = results_graph.query(distinct_messages)
for row in messages:
    print(row.message)

#each sh:resultMessage in the results_graph once, along with the number of times that message has been repeated in the results
count_messages = """
PREFIX sh: <http://www.w3.org/ns/shacl#> 

SELECT ?message (COUNT(?node) AS ?num_messages) WHERE {
    [] sh:result ?errorBlankNode .
    ?errorBlankNode sh:resultMessage ?message ;
                    sh:focusNode ?node .
}
GROUP BY ?message
ORDER BY DESC(?count) ?message
"""

messages = results_graph.query(count_messages)
for row in messages:
    print(f"COUNT: {row.num_messages} | MESSAGE: {row.message}")

RDFS (Lab 7)

import owlrl
from rdflib import Graph, RDF, Namespace, FOAF, RDFS

g = Graph()
ex = Namespace('http://example.org/')

g.bind("ex", ex)
g.bind("foaf", FOAF)

NS = {
    'ex': ex,
    'rdf': RDF,
    'rdfs': RDFS,
    'foaf': FOAF,
}

#Write a small function that computes the RDFS closure on your graph.
def flush():
    owlrl.DeductiveClosure(owlrl.RDFS_Semantics).expand(g)

#Rick Gates was charged with money laundering and tax evasion.
g.add((ex.Rick_Gates, ex.chargedWith, ex.MoneyLaundering))
g.add((ex.Rick_Gates, ex.chargedWith, ex.TaxEvasion))

#When one thing that is charged with another thing,
g.add((ex.chargedWith, RDFS.domain, ex.PersonUnderInvestigation))  #the first thing is a person under investigation and
g.add((ex.chargedWith, RDFS.range, ex.Offense))  #the second thing is an offense.

#Write a SPARQL query that checks the RDF type(s) of Rick Gates and money laundering in your RDF graph.
print('Is Rick Gates a ex:PersonUnderInvestigation:', g.query('ASK {ex:Rick_Gates rdf:type ex:PersonUnderInvestigation}', initNs=NS).askAnswer)
print('Is Money Laundering a ex:Offense:', g.query('ASK {ex:MoneyLaundering rdf:type ex:Offense}', initNs=NS).askAnswer)
flush()
print('Is Rick Gates a ex:PersonUnderInvestigation:', g.query('ASK {ex:Rick_Gates rdf:type ex:PersonUnderInvestigation}', initNs=NS).askAnswer)
print('Is Money Laundering a ex:Offense:', g.query('ASK {ex:MoneyLaundering rdf:type ex:Offense}', initNs=NS).askAnswer)

#A person under investigation is a FOAF person
g.add((ex.PersonUnderInvestigation, RDFS.subClassOf, FOAF.Person))
print('Is Rick Gates a foaf:Person:', g.query('ASK {ex:Rick_Gates rdf:type foaf:Person}', initNs=NS).askAnswer)
flush()
print('Is Rick Gates a foaf:Person:', g.query('ASK {ex:Rick_Gates rdf:type foaf:Person}', initNs=NS).askAnswer)

#Paul Manafort was convicted for tax evasion.
g.add((ex.Paul_Manafort, ex.convictedFor, ex.TaxEvasion))
#the first thing is also charged with the second thing
g.add((ex.convictedFor, RDFS.subPropertyOf, ex.chargedWith)) 
print('Is Paul Manafort charged with Tax Evasion:', g.query('ASK {ex:Paul_Manafort ex:chargedWith ex:TaxEvasion}', initNs=NS).askAnswer)
flush()
print('Is Paul Manafort charged with Tax Evasion:', g.query('ASK {ex:Paul_Manafort ex:chargedWith ex:TaxEvasion}', initNs=NS).askAnswer)

print(g.serialize())