Hands-On Project | Build a Movie Knowledge Graph

Phase I — Domain Analysis

Before building, understand the movie domain thoroughly.

Learning Objectives

Analyze movie domain entities
Identify classes and properties
Explore existing vocabularies
Map concepts to RDF

Domain Entities

Entities

Movie, Person, Director, Actor, Genre, Studio

Relationships

directedBy, starring, hasGenre, producedBy

Attributes

title, releaseDate, runtime, budget, rating

Vocabularies

schema.org, DBpedia, FOAF, Wikidata

Why Reuse Vocabularies?

Enables interoperability with other datasets
Provides established semantics
Allows federated queries
Follows Linked Data best practices

Schema.org/Movie DBpedia Ontology FOAF Spec

Exercise — Domain Mapping

A movie can have multiple actors — what property type is this?
A director directs movies — should this be functional?
How would you model the "sequel of" relationship?

Phase II — Design Ontology

Create a formal ontology using RDFS and OWL.

Complete the Ontology

Add :Director and :Actor as subclasses of :Person
Add :Genre class
Define :directedBy, :starring, :hasGenre properties
Add domain/range constraints
Make :directedBy functional with inverse

Ontology Editor

Turtle

Phase III — Query DBpedia

Query real Linked Data from DBpedia.

About DBpedia

DBpedia extracts structured data from Wikipedia. It contains millions of entities including movies, people, and places — all available as Linked Data.

DBpedia SPARQL Endpoint DBpedia Ontology

SPARQL Queries

SPARQL

Open in DBpedia →

Exercise

Find all movies released in 2023
Get the top 10 highest-budget films
Find movies sharing actors with "Inception"

Phase IV — Build Dataset

Create instance data and link to external sources.

Linking to External Data

Use owl:sameAs to link your entities to DBpedia and Wikidata for seamless data integration across the Linked Data cloud.

Movie Dataset

Turtle

Extend the Dataset

Add 2 more movies with all properties
Add a new actor who appears in multiple films
Add owl:sameAs links to DBpedia

Phase V — Reasoning

Understand how reasoners derive new knowledge.

RDFS Inference Rules

Subclass Transitivity

If A ⊆ B and B ⊆ C, then A ⊆ C

:Director rdfs:subClassOf :Person .
:nolan a :Director .
# → :nolan a :Person

Domain Inference

If P has domain C, x P y → x ∈ C

:directedBy rdfs:domain :Movie .
:inception :directedBy :nolan .
# → :inception a :Movie

Range Inference

If P has range C, x P y → y ∈ C

:directedBy rdfs:range :Director .
:inception :directedBy :nolan .
# → :nolan a :Director

Inverse Properties

If P inverseOf Q, x P y → y Q x

:directedBy owl:inverseOf :directed .
:inception :directedBy :nolan .
# → :nolan :directed :inception

Open World Assumption

In Semantic Web, missing data is unknown, not false. This differs fundamentally from traditional databases.

Missing information ≠ Non-existence
Use owl:complementOf for explicit negation
Use SPARQL NOT EXISTS for closed-world queries

Inference Examples

Asserted Fact	Rule	Inferred Fact
`:nolan a :Director`	rdfs:subClassOf	`:nolan a :Person`
`:inception :directedBy :nolan`	owl:inverseOf	`:nolan :directed :inception`
`:inception :starring :dicaprio`	rdfs:range	`:dicaprio a :Actor`

Reasoning Practice

What's inferred if :Studio rdfs:subClassOf :Organization?
If :hasSequel is transitive, and A hasSequel B, B hasSequel C — what's inferred?
How would you detect someone who is both a Director and Actor?

Phase VI — SPARQL Analytics

Write advanced analytical SPARQL queries.

Analytics Queries

SPARQL

SPARQL Analytics Features

Aggregates

COUNT, SUM, AVG, MIN, MAX, GROUP_CONCAT

Grouping

GROUP BY, HAVING for filtering groups

Subqueries

Nested SELECT for complex logic

Path Queries

Property paths: *, +, ?, |, / operators

Write Queries

Find the director with the highest average movie rating
List actors who have worked with multiple directors
Calculate total budget per genre per year

Phase VII — Visualization

Build an interactive knowledge graph visualization.

Knowledge Graph

Movie

Director

Actor

Genre

Congratulations!

You've completed the hands-on project. You now have practical experience with:

Designing ontologies with RDFS and OWL
Querying real Linked Data from DBpedia
Creating and validating RDF datasets
Understanding semantic reasoning and inference
Writing analytical SPARQL queries
Visualizing knowledge graphs

Next Steps & Resources

Protégé (OWL Editor) Apache Jena RDFLib (Python) GraphDB