Development of a Movies Ontology and a Semantic Web Application that Suggests Movies Based on User's Favor Criteria

1
Development of a Movies Ontology and a Semantic
Web Application that Suggests Movies Based on
User's Favor Criteria
Yüce TEKOL
2
Overview

• Introduction
• IMDB Ontology
• Cwmovies
• Test Database
• Test 1
• Test 2
• Test 3

3
Introduction
The Internet Movie Database (http//www.imdb.com)
lists details of millions of movies, tv shows,
etc. with many detatils such as the director, the
cast and genre for cinema fans. Being a vast
database, it presents a good opportunity to study
and experiment Semantic Web concepts also,
developed tools may be useful for many people.
Cwmovies is a Python application that recommends
movies based on a user's favor criteria. Cwmovies
uses Tim Berner's Lee's Cwm as its reasoning
engine.
4
IMDB Ontology
The ontology used by Cwmovies is designed in
SWOOP. The ontology imports the IMDB ontology
written by Gunnar Aastrand Grimnes, and adds
genre property which is missing from it.
5
IMDB Ontology (cont'd)
Below is the ontology used by Cwmovies.
6
Cwmovies
Cwmovies is invoked from the command line
together with the full path name of a file which
contains the user's favor criteria, such as the
movie should be a science-fiction movie and its
director should be Steven Spielberg written in
N3 query format. Cwmovies has two
functions 1.Builds the movies database which is
in RDF/XML by fetching information from
IMDB. 2.Runs Cwm on the database and the user
favor criteria file and interprets the outputs,
finding the movies that correspond to the
criteria.
7
Cwmovies - Screenshot
8
Cwmovies - Tools
Following tools have been used for the
development of Cwmovies 1.Python 2.4.3
lthttp//www.python.orggt 2.IMDbPy 2.8
lthttp//imdbpy.sourceforge.netgt 3.rdflib 2.3.3
lthttp//rdflib.netgt 4.Pawk 0.0.1
lthttp//code.google.com/p/py-pawk/gt
9
Test Database
All the tests run on an RDF/XML file which is
automatically fetched and converted by Cwmovies
and contains title, year, director and genre
information of 153 movies which are in IMDB's top
movies of the following genres Crime, Thriller,
Horror and Science-fiction.
10
Test Database (cont'd)
11
Test 1
Favor criteria
All thriller movies produced after 2000
N3 Query
_at_prefix dc lthttp//purl.org/dc/elements/1.1/gt
. _at_prefix imdb lthttp//purl.org/NET/tekol/yuce/20
07/imdb-nsgt . _at_prefix cwmv lthttp//purl.org/NET/
tekol/yuce/2007/cwmovies-nsgt . _at_prefix math
lthttp//www.w3.org/2000/10/swap/mathgt . ?x
imdbgenre "Thriller". ?x.imdbyear
mathgreaterThan 2000. gt ?x a
cwmvfavorite .
12
Test 1 (cont'd)

• Batman Begins
• Children of Men
• Donnie Darko
• El Laberinto del Fauno
• Hotel Rwanda
• Kill Bill Vol. 1
• Kill Bill Vol. 2
• Minority Report
• Serenity
• Sin City
• Spider-Man 2
• The Departed
• The Others
• V for Vendetta
• X2

Results
13
Test 2
Favor criteria
Any of Francis Ford Coppola's Films
N3 Query
_at_prefix dc lthttp//purl.org/dc/elements/1.1/gt
. _at_prefix imdb lthttp//purl.org/NET/tekol/yuce/20
07/imdb-nsgt . _at_prefix cwmv lthttp//purl.org/NET/
tekol/yuce/2007/cwmovies-nsgt . ?x
imdbdirector "Francis Ford Coppola" gt ?x a
cwmvfavorite .

• The Conversation
• The Godfather
• The Godfather Part II

Results
14
Test 3
Favor criteria
Crime movies produced between 1960 and 1975
N3 Query
_at_prefix dc lthttp//purl.org/dc/elements/1.1/gt
. _at_prefix imdb lthttp//purl.org/NET/tekol/yuce/20
07/imdb-nsgt . _at_prefix cwmv lthttp//purl.org/NET/
tekol/yuce/2007/cwmovies-nsgt . _at_prefix math
lthttp//www.w3.org/2000/10/swap/mathgt . ?x
imdbgenre "Crime". ?x.imdbyear
mathgreaterThan 1960. ?x.imdbyear
mathlessThan 1975. gt ?x a cwmvfavorite .
15
Test 3 (cont'd)

• A Clockwork Orange
• Butch Cassidy and the Sundance Kid
• Cape Fear
• Chinatown
• In Cold Blood
• Le Samouraï
• The Conversation
• The Godfather
• The Godfather Part II
• The Sting
• Yojimbo

Results
16
Conclusions
A Semantic Web application that makes use of vast
movies database of IMDb together with an ontology
is presented. The system is capable of presenting
movies which conform to user's favor
criteria. Currently, only genre, title, year and
director fields can be included in the query, but
adding other fields is not hard.
17
That's All...