====== Modeling and using knowledge in RDF ======

^  Last verification: | 20180912 |
^  Tools required for this lab: | -- |

===== Before the lab =====
Reading:
  * [[http://www.w3.org/TR/rdf-schema/|RDF Schema 1.1]]
  * [[#if_you_want_to_know_more|If you want to know more...]]

===== Lab instructions =====

<WRAP center round important 100%>
During this lab we will use [[http://any23.org/|Anything to Triples]] website. If it is down, there is also a possibility to download any23 service and run it on your own computer:
  * download **Apache Any23 Service (Standalone server embedded) 1.1** file from https://any23.apache.org/download.html (mirror: {{http://krzysztof.kutt.pl/didactics/semweb/apache-any23-service-1.1-server-embedded.tar.gz|apache-any23-service-1.1-server-embedded.tar.gz}}) and unpack the zip somewhere
  * execute command to start server with any23 service: <code>[PATH-to-unpacked-archive]/apache-any23-service-1.1-server-embedded/bin/any23server </code>
    * Note: use the any23server.bat file if you are using Windows OS
    * Change ''<nowiki>"$REPO"</nowiki>'' into ''%REPO%'' twice (line #79 in any23server.bat file) if ''Error: Unable to access jarfile $REPO/jetty-runner-8.1.4.v20120524.jar.'' message appears
  * after few seconds any23 service will appear on [[http://localhost:8080/apache-any23-service/]] address :)
</WRAP>

During this lab, you will serialize (using Turtle syntax) an RDF graph created before (The Bold and the Beautiful vs The Game of Thrones).

==== - Modeling knowledge with RDF triples [20 minutes] ====
RDF is a data model for which several syntaxes have been developed.
RDF document is an RDF graph (describing some objects) serialized into chosen representation/syntax.
In this task, you will convert the graph representation into so called Turtle syntax.

  - Create a new document with a ''*.ttl'' extension.
  - Create a document describing the graph created during the [[lab-rdfmodel#modeling_knowledge_with_rdf_graphs_25_minutes|previous lab]]. Use the [[http://www.w3.org/TR/turtle/|RDF Turtle]] for reference.
    - Define namespaces for your document URIs, e.g., a standard RDF namespace and one "your own" for this specific topic:  <code>@prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>.
@prefix bb: <http://yourname/b-and-b#>.</code>
      * You can also use a base namespace: <code>@base   <http://kkutt/b-and-b/> .</code>
    - Write down ~14 selected triples from your graph. Make sure you include each of the following at least once:
      - resources, 
      - datatype values.
    - Notes:
      * Comments in Turtle begins with ''#'' symbol.
      * In RDF we are describing something (a subject) and giving it a unique ID, e.g.: ''<nowiki><http://kkutt/b-and-b/brooke-logan></nowiki>''. To make notation more compact, we can use [[http://www.w3.org/TR/turtle/#prefixed-name|prefixes]] and/or [[http://www.w3.org/TR/turtle/#relative-iri|base namespace]].
      * 8-) To separate statements we can use dot (.), semicolon (;) and comma (,). What is the difference between them?
    - Go to the <wrap caution>[[http://any23.org/|Anything to Triples]]</wrap> webpage:
      * Validate your document using this tool (section "Convert copy&pasted document"):
        * input format: **Turtle (text/turtle)**
        * output format: **turtle**
        * validation: **validate**
        * report: **checked**
        * annotate: **unchecked**
        * paste your code and click "Convert" button
      * if something is wrong, you can see details in report (in web browser), e.g.: \\ {{:pl:dydaktyka:semweb:any23_some_problems_web.png?400|}}
      * if code was successfully validated then you get an XML with a lot of empty tags and your Turtle code inside ''<data><![CDATA[ ... ]]></data>'' (indentation may be changed)
    - 8-) Include model (''.ttl'' file) in the report archive.

==== - RDF: Containers and Collections [10 minutes] ====
In RDF there are two ways to describe set or sequences of objects: //[[http://www.w3.org/TR/rdf11-mt/#rdf-containers|Containters]]// and //[[http://www.w3.org/TR/rdf11-mt/#rdf-collections|Collections]]//.

The container and collection are resources that contain things. The contained things are members.
RDF defines three types of containers:
    * ''rdf:Bag'',
    * ''rdf:Seq'',
    * ''rdf:Alt'',
and one type of collection:
    * ''rdf:List''.
While "a //container// only says that certain identified resources are members; it does not say that other members do not exist." with a //Collection// we can describe groups containing __only__ the specified members.

[[http://www.w3.org/TR/rdf11-mt/#rdf-containers|Read about them]] and then:
  - 8-) Describe shortly what is the difference between different types of containers.
  - Create a container in your database.
  - In your RDF file create also an //RDF Collection//. For hint and examples see the [[http://www.w3.org/TR/turtle/#collections|recommendation]].

==== - RDF: Datatypes [10 minutes] ====
Add the references to [[http://www.w3.org/TR/xmlschema-2/|XML Schema datatypes]] to chosen information in your RDF file, e.g.:
<code>
# ...
@prefix xsd: <http://www.w3.org/2001/XMLSchema#> .
# ...

# ...
<brooke-logan> <daughter-of> <beth-logian>;
	<born> "1900-01-01"^^xsd:date;
	<studied> "Chemistry"^^xsd:string;
# ...
</code>
Use at least 3 different datatypes!

==== - RDF Visualization [3 minutes] ====

We are in the middle, so let's visualize our RDF files.

  - There are two ways to do this:
    - Using <wrap caution>[[http://any23.org/|Anything to Triples]]</wrap> service convert your file to RDF/XML (''rdfxml'') format. Do not analyze resulting code. Simply copy it :) \\ Go to the [[http://www.w3.org/RDF/Validator/|RDF Validator]] page and simply paste generated code into ''Check by Direct Input'' field. Select ''Graph only'' from dropdown list and click ''Parse RDF''
    - Use the [[http://loki.ia.agh.edu.pl/wiki/docs:rdfeditor|RDF Editor]] developed at AGH UST (by Artur Smaroń, EIS 2015-2016)
  - Analyze the results:
    - 8-) How containers are represented in the graph?
    - 8-) How collections are represented in the graph?

==== - Semantic vocabularies: FOAF, Dublin Core and FHKB [10 minutes] ====

Semantic vocabularies are sets of predefined properties for describing some domains. Examples include:
  * [[http://semanticweb.org/wiki/Dublin_Core|Dublin Core]]
  * [[http://semanticweb.org/wiki/FOAF|FOAF]]
  * [[http://semanticweb.org/wiki/SIOC|SIOC]]
  * etc.

  - Have a look at [[https://github.com/blokhin/genealogical-trees/blob/master/data/header.ttl|Family History Knowledge Base (FHKB)]] ontology, [[http://dublincore.org/documents/dc-rdf/|Expressing Dublin Core metadata using the Resource Description Framework]] and [[http://xmlns.com/foaf/spec/|FOAF Vocabulary Specification]].
  - Modify your RDF file to use the properties from FHKB, Dublin Core and FOAF vocabulary, e.g.:
<code>
# ...
@prefix dc: <http://purl.org/dc/elements/1.1/>.
@prefix dcterms: <http://purl.org/dc/terms/> .
@prefix foaf: <http://xmlns.com/foaf/0.1/> .
@prefix fhkb: <http://www.example.com/genealogy.owl#> .
# ...

# ...
<brooke-logan> fhkb:isDaughterOf <beth-logian>;
	<born> "1900-01-01"^^xsd:date;
	<studied> "Chemistry"^^xsd:string;
	foaf:name "Brooke Logan", "Brooke Forrester", "Brooke Chambers", "Brooke Jones", "Brooke Marone", "Brooke Spencer".
# ...
</code>
    * Hint: comprehensive list of ''dcterms:'' and ''dc:'' elements is [[http://dublincore.org/documents/dcmi-terms/|here]] ("Section 2: Properties in the /terms/ namespace" = ''dcterms:'', "Properties in the /elements/1.1/ namespace" = ''dc:'')


==== - RDF Schema: classes [10 minutes] ====
RDF Schema allows to organize objects into classes and define simple taxonomies.

  - Define classes of items in your database. Consider reusing the classes from dictionaries mentioned before, e.g.
<code>
@base   <http://kkutt/got/> .
@prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> .
@prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> .

<Character> rdf:type rdfs:Class .

<Dwarf> rdf:type rdfs:Class ;
	rdfs:subClassOf <Character> .

foaf:Person rdfs:subClassOf <Character> .
</code>

Add the ''rdf:type'' statements to your RDF file, e.g.:
<code>
<brooke-logan> a foaf:Person .
<tyrion-lannister> rdf:type <Dwarf> .
</code>

Note that ''a'' is an alias for ''rdf:type'' so two statements: <code><tyrion-lannister> rdf:type <Dwarf> .</code><code><tyrion-lannister> a <Dwarf> .</code> are equivalent.

==== - RDF Schema: properties [10 minutes] ====

RDF Schema also provides a way to define Properties as well as their domains and ranges.

  - In your file identify properties that are not defined in external dictionaries (e.g. Dublin Core), e.g. in this code:<code>
<brooke-logan> fhkb:isDaughterOf <beth-logian>;
	<born> "1900-01-01"^^xsd:date;
	<studied> "Chemistry"^^xsd:string;
	foaf:name "Brooke Logan", "Brooke Forrester", "Brooke Chambers", "Brooke Jones", "Brooke Marone", "Brooke Spencer".
</code> ''<studied>'' is not defined in external dictionary -- it is defined in our own namespace.
  - Describe each such property using RDF Schema capabilities:<code>
@base   <http://kkutt/b-and-b/> .
@prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> .
@prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> .

<studied> rdf:type rdfs:Property ;
	rdfs:domain foaf:person ;
	rdfs:range rdfs:Literal.
</code>

You may find [[http://www.w3.org/TR/rdf-schema/|RDF Schema 1.1]] recommendation useful.

==== - RDF Schema: Non-modeling properties [3 minutes] ====

RDF Schema also provides some handful properties that are not used in inference process. These are:
  * ''rdfs:label'' -- used by convention to provide a human-readable name that is displayed by semantic web agents,
  * ''rdfs:seeAlso'' -- cross-referencing; provide links to supplementary information,
  * ''rdfs:isDefinedBy'' -- subPropertyOf rdfs:seeAlso; provides a link to the primary source of information,
  * ''rdfs:comment'' -- for everything you want :) .

  - Use each of them in your RDF file.
  - 8-) Include final RDF **Turtle** file in the report archive.
    * Check if your Turtle file passes validation on <wrap caution>[[http://any23.org/|Anything to Triples]]</wrap>!
===== Control questions =====
   * What container elements are available in RDF?
   * What is RDF Schema?
   * What are core RDF Schema Classes and properties?
   * How constraints on domain and range of properties are added?

===== If you want to know more =====

{{section>.:lab-rdfmodel#if_you_want_to_know_more&noheader}}