This paper presents a web service which delivers personalized audio information. The personalization process is automated and decentralized. The metadata which support personalization are separated in two categories: the metadata describing user preferences stored at each user and the resource adaptation metadata stored at the web service host. The multimedia models MPEG-21 and MPEG-7 are used to describe metadata information and the Web Ontology Language (OWL) to produce and manipulate ontological descriptions. SPARQL is used for querying the OWL ontologies. The MPEG Query Format (MPQF) is also used, providing a wellknown framework for applying queries to the metadata and to the ontologies.

1. 1
A Personalized Audio Web Service using
MPEG-7 and MPEG-21 standards
Emmanouil Skondras1, Malamati Louta2,
Panagiotis Sarigiannidis2
1Technological Educational Institute of Western Macedonia
2University of Western Macedonia
PCI 2011

2. 2
Introduction
• The application uses MPEG-7 and MPEG-21 for
the description of audio content as well as users’
preferences, which are the basic metadata for the
semantic description of multimedia content.
• The metadata information is managed using Web
Ontology Language (OWL) ontologies.
PCI 2011

3. 3
Related Work Examples (1/2)
• [Kohncke] presents an adaptation model for
content personalization by integrating MPEG-
7/21 metadata.
• It uses web services as basic modules.
• A central web service monitors a suitable
workflow in respect of user preferences, content
semantics, network constrains and terminal
capabilities.
PCI 2011

4. 4
Related Work Examples (1/2)
• Each web service evaluates the MPEG-7/21
description and adapts the multimedia
material.
• The user gets the best possible quality in
respect of his terminal specifications.
PCI 2011

5. 5
Related Work Examples (2/2)
• The work presented in [Zhai] pays attention to
semantic retrieval for sports information in the
World Wide Web.
• SPARQL is used for queries execution.
• Intelligent retrieval is realized according to
relations between sports such as “synonymy
of”, “kind of” and “part of”.
PCI 2011

6. 6
Related Work Examples (2/2)
• The process is as follows:
– A sports-ontology is created.
– Data are collected from data sources and annotated
with the ontology.
– The search engine completes semantic matching of
retrieval conditions through ontology reasoning for
user's request and finds out the eligible data.
PCI 2011

7. 7
Used Standards
• MPEG-7
– a multimedia content description standard
• MPEG-21
– defines a framework for adaptive multimedia
applications
• OWL
– is adopted so as to create the relative ontologies and
provide a common semantic understanding between
the components involved in the personalization
process
PCI 2011

8. 8
Software Architecture
• The basic modules of our architecture:
• The main web service proposes audio tracks to the
clients, based on their choices and preferences.
PCI 2011

9. 9
Software Architecture
• The main web server contains the audio tracks
and the respective audio metadata using MPEG-
7 in an MPEG-21 structure.
• The audio tracks are divided in sixteen different
audio categories (pop, classical etc.).
PCI 2011

10. 10
Software Architecture
• Audio metadata include:
– User defined metadata
– Technical oriented metadata
– Usage history metadata
• track’s popularity in respect to all tracks
• track’s popularity in its category
• The relative OWL ontology is used for the audio
metadata manipulation.
PCI 2011

11. 11
Software Architecture
• The client’s metadata rely on user’s preferences
– Favorite audio categories
– Top 10 audio tracks in each category
• The relevant OWL ontology is used for the user
preferences manipulation.
PCI 2011

12. 12
Software Architecture
• The architecture is decentralized.
– Each client manipulates its own metadata locally.
– Client’s metadata are organized according to the
MPEG-21/7 user preferences element of the MPEG-
21 usage environment.
PCI 2011

13. 13
Software Architecture
• The main web service and the client
communicate with:
– MPQF parser web service
• Interacts with MPEG-21/7 metadata through MPQF queries
– SPARQL parser web service
• Interacts with OWL ontologies through SPARQL queries
PCI 2011

14. 14
• The client uploads user defined metadata together
with the uploaded track.
• The server uses the Java Media Framework to
analyze the uploaded audio track and extracts
technical oriented metadata.
Audio Metadata Creation
PCI 2011

15. 15PCI 2011
Audio Metadata Creation

16. 16
Audio Promotion
PCI 2011

17. 17
Weighting formula
Weight values are updated according to the formula:
• pref_valuei : the preference value of the requested
audio file i.
• wp+wh=1.

 N
j
j
i
hphp
valuepref
valuepref
ww
1
//
_
_
PCI 2011

18. 18
Weighting formula
• When the user requests to listen to an audio file
that has been promoted according to the user
preferences, wp increases while wh decreases.
• Adversely, the opposite operation is performed
when the audio file has been promoted according
to the usage history values.
PCI 2011

19. 19
Case Study
• The client contacts the main web service and sends
his preferences through the relative OWL
ontology.
• The client retrieves an audio catalog using the
SPARQL query:
PREFIX mpeg7: < http://www.mpeg.org/MPEG7/2000>
SELECT ?Title
WHERE { ?x mpeg7:title ?title .
FILTER (?Genre=Classical || ?Genre=Jazz && ?Publisher=Sony &&
?CreationDate>=2005-01-01 && ?MediaDuration>60 &&
?Format=MP3 && ?FileSize>10000 && ?BitRate>=42000)
}ORDER BY DESC(?preferenceValue)
PCI 2011

20. 20
Case Study
• Consequently the client requests a specific
audio track.
• The main web service uses the relative OWL
ontology, manipulates the audio metadata and
sends the requested track to the client.
PCI 2011

21. 21
Case Study
• Response times for audio track and SPARQL
request in respect of tracks’ number
PCI 2011

22. 22PCI 2011