An Architecture based on Ontologies, Agents and Metaheuristics
Applied to the Multimedia Service of the Brazilian Digital
Television System
Toni Ismael Wickert and Arthur Tórgo Gómez
Universidade do Vale do Rio dos Sinos, Av. Unisinos - 950, São Leopoldo, Brazil
Keywords: Ontologies, Metaheuristics, Tabu Search, Genetic Algorithm, Software Agents.
Abstract: With the advent of the Brazilian Digital Television System, that arrives on approximately 95% of Brazilian
homes, the users will be able to have an interactive channel by the utilization of the digital television. Thus,
will be possible to access the multimedia application server, i.e., to send or to receive emails, to access
interactive applications, to watch movies or specific news. This paper proposes the development and the
implementation of an architecture that includes a module that suggests the content to the user according to
his profile and another module to optimize the content that will be transmitted. The implementation was
developed using ontologies, software agents, Tabu Search and Genetic Algorithm. The validations of the
results are done using a metric.
1 INTRODUCTION
In Brazil is occurring a technological evolution with
the advent of digital television, for example, the
signal that was received by viewers in analog way
before, now is received in digital way, which
improves the image quality avoiding the appearance
of the flickering screen and of the images ghosting,
commonly found in analog transmissions (Carvalho,
2006).
Nowadays, in Brazil, the broadcast television is
the most important access to information.
Considering the 58.4 million Brazilian households,
55.4 million have at least one receiver. In other
words, the public television reaches 95% of
Brazilian Households (Bip, 2001).
Beyond the quality improvements and the image
transmission, the Brazilian Digital Television
System includes the interactive channel that allows
the interactive applications access by users. The
station may have a server with multimedia
applications and the user can access them through
the interactive channel. In this context, the return
channel can be used for data transmission such as
emails, interactive questionnaires and multimedia
content like interactive distance courses,
presentations with interact content, among others
(Manhães, 2005).
The purpose of this work is the implementation
of a system divided into two modules: content
suggestion module and content transmission module.
In the content suggestion module will be used
software agents and ontologies in order to classify
content crossing it with the user profile suggesting
the appropriate content. In the content transmission
module will be used software agents that with the
Tabu Search (TS) and the Genetic Algorithm (GA)
will define the better transmission policy, through
the optimizing of the transmission parameters, such
as transmission rates of audio and video, among
others.
This paper is organized as follows. Section 2
presents the related work. Section 3 presents the
proposed architecture. Section 4 shows the results
obtained. And Section 5 presents the conclusions.
2 SIMILAR WORKS
In this section will be presented similar studies that
used each of the technologies that are involved in
this work. Will be presented related works that use
ontologies to suggest content for Digital TV with
software agents and other works that use
metaheuristics to the transmission parameter
optimization. In Literature aren’t found papers that
203
Wickert T. and Tórgo Gómez A..
An Architecture based on Ontologies, Agents and Metaheuristics Applied to the Multimedia Service of the Brazilian Digital Television System.
DOI: 10.5220/0004001402030208
In Proceedings of the International Conference on Data Technologies and Applications (DATA-2012), pages 203-208
ISBN: 978-989-8565-18-1
Copyright
c
2012 SCITEPRESS (Science and Technology Publications, Lda.)
addressing in a unique work all the technologies
applied in this work. Also, were not found works in
the literature that optimize the same types of
parameters considered in this work and therefore it
is hard to compare it to existing works.
2.1 Metaheuristics
The metaheuristics used in this work are the Tabu
Search, proposed by Fred Glover in 1986 (Glover,
1986), and the Genetic Algorithm proposed by John
Holland in 1975 (Holland, 1975).
The metaheuristics have been used in a variety of
combinatorial optimization problems. Gendreau,
Laporte and Potvin (2002) and Simas (2007) used
metaheuristics for the Vehicle Routing Problem.
Since Chung, et al. (2010), Gonçalvez and Mauricio
(2004), Gonçalvez and Tiberti (2006) make use of
metaheuristics applied to the Manufacturing Cells
Formation Problem.
The utilization of metaheuristics for parameter
optimization of data transmission for IPTV was
approached by Weissheimer (2011), in this paper the
author presents the development of a computational
model based on the application of metaheuristics on
an IPTV platform in order to find the best
configuration of transmission parameters
considering the types of users and receiving devices.
Link (2011) presents a system to configure
parameters of video coding for digital TV using the
H.264 standard. This search is based on
metaheuristics Tabu Search and Genetic Algorithms.
In this work was developed a hybrid algorithm,
based on the utilization of these two metaheuristics,
the Tabu Search was used to intensify the search in
conjunction with the power of GA diversification.
2.2 Software Agents and Ontologies
In the context of digital television have been
published several works using ontologies for the
representation of programs, movies and sports
metadata available to users.
In the paper of Araújo and Ricarte (2010) the
authors proposed an integration of existing metadata
in a transmission environment and reception of
digital TV in open networks of terrestrial and
satellite transmission. Once the open digital TV
receiver device will have access to an interactive
channel via Internet, was presented a methodology
to integrate the metadata information of the
broadcast industry using ontologies to describe the
knowledge of specific areas in existing ontology
repositories on Internet.
Fernández, et al. (2006), proposed an automatic
content recommendation system for Digital TV
broadcast programs based on the Web Semantic
technologies like ontologies, OWL (Web Ontology
Language) and software agents.
Figure 1 shows the ontology structure that was
set up, can observe a similar structure to the
ontology propused in this work, however, with a
larger scope, because the authors encompassed
different contents such as movies, series,
entertainment, concerts, among others.
Figure 1: AVATAR ontology structure.
3 PROPOSED SOLUTION
The Brazilian Digital Television System supports an
interactive channel, considering that exist an internet
connection to make the communication with the
television station. From this fact, were proposed two
ontologies. The first one, that aims to make a
classification of content in the multimedia
application server, for example, classification of
films by genre, film's authors, release year, etc. and
other ontology to describe the users profile, for
example, can be identified if the users that like to
watch movies of action genre, comedy, suspense,
among others.
Thus, with these two ontologies, it is possible to
make a crossing of the user profile with the content
available on the multimedia application server. To
do it was created a software agent that reads the
user's profile, to find the content available on the
multimedia application server and to send this
suggestion to the user. These suggestions will be
sent in a time interval which can be determined by
the television station.
After the user accepts the suggested content, a
communication channel is established between the
multimedia application server and the user. On the
other hand, the optimizer agent, that is processed
from time to time on the server, will determine the
DATA2012-InternationalConferenceonDataTechnologiesandApplications
204
most optimum way to deliver the content to end
user. This agent will run a metaheuristic that will
determine the parameters and the best policy for the
transmission according to the profile defined by
ontologies.
Figure 2 shows a practical example of the
operation of the proposed architecture. At the
content suggestion module there are the consultants
agents, one for each user, they are responsible for
analyzing user's profile, in this case we have two
users, each one with a profile described by an
ontology. After knowing the profile, each agent
performs a search in the server's ontology to see the
contents that are appropriate for each user. Suppose
that user likes to watch action movies, the agent will
search for contents related to action movies.
Figure 2: Proposed architecture.
After the acceptance of a suggested content, the
control passes to the transmission content module. In
this module we have the coordinator agent and the
optimizers agents. The coordinating agent will
trigger the other agents to run the metaheuristics to
determine the best parameters and the best policy for
transmission of the content. After ended this process,
each optimizer agent will transmit the results found
by the metaheuristic Tabu Search and Genetic
Algorithm for the coordinating agent. The
coordinating agent will analyze which of them had
the best result. These parameters will be chosen for
the transmission of content between the multimedia
application server and the user. In this procedure, is
considered that content streaming will occur over the
IPTV. To finish the transmission the communication
channel is closed.
3.1 Multimedia Application Server and
User's Ontology
The multimedia application server ontology was
created to describe the multimedia content that is
stored on the server. Due to the great diversity of
content that a server can store this work will restrict
the scope of this ontology for movies.
The user's ontology was created to describe the
user's profile that connects to the server. Due to the
large amount of data that can be described on a
user's profile, the scope was limited to some relevant
information.
Figure 3 shows the multimedia application
server's ontology together with the user's ontology.
The light circles indicate the ontology classes and
the dark circles indicate classes instances.
Figure 3: Ontology structure.
3.2 Mathematical Formulation
This section presents the objective function (OF) and
restrictions related to the OF. For each decision
variable of the OF will be presented the range of
feasible values. The expected result of the
formulation is to have the best possible utilization
and distribution of bandwidth of the server in order
to attend a varying number of clients connected to
the service.
(1)
Restrictions:
0.01 ≤ vLD ≤ 1.00 (2)
1.00 ≤ vP1 ≤ 2.00 (3)
AnArchitecturebasedonOntologies,AgentsandMetaheuristicsAppliedtotheMultimediaServiceoftheBrazilianDigital
TelevisionSystem
205
2.00 ≤ vSD ≤ 5.00 (4)
5.00 ≤ vP2 ≤ 10.00 (5)
10.00 ≤ vHD ≤ 18.00 (6)
0.096 ≤ aST ≤ 0.256 (7)
0.384 ≤ a51 ≤ 1.00 (8)
LB = 500.00 (9)
α + β + γ + δ + ω + θ + ρ ≥ 0 (10)
nLD, nSD, nHD, nP1, nP2, nST, n51 ∈ Z
+
(11)
(12)
Where:
• LB = total bandwidth of the television station;
• vLD = LD (low definition) video quality;
• vP1 = P1 (between LD and SD definition) video
quality;
• vSD = SD (standard definition) video quality;
• vP2 = P2 (between SD and HD definition) video
quality;
• vHD = HD (high definition) video quality;
• aST = Stereo audio quality;
• a51 = multichannel 5.1 audio quality;
• nLD = number of clients connected as LD;
• nSD = number of clients connected as SD;
• nHD = number of clients connected as HD;
• nP1 = number of clients connected as P1;
• nP2 = number of clients connected as P2;
• nST = number of clients connected as Stereo audio
quality;
• n51 = number of clients connected as multichannel
5.1 audio quality;
• α = importance level of the LD transmission;
• β = importance level of the SD transmission;
• γ = importance level of the HD transmission;
• δ = importance level of the P1 transmission;
• ω = importance level of the P2 transmission;
• θ = importance level of the Stereo Audio
transmission;
• ρ = importance level of the multichannel 5.1
transmission;
• All decision variables accept real numbers.
It can be seen, through the mathematical formulation
of the equations, that the system bottleneck is the
bandwidth available at the multimedia application
server. Because this feature is limited and has a high
cost, it should be used as best as possible.
4 EXPERIMENTS AND RESULTS
The validation of the experiments for the content
suggestion module was made through simulations
and analysis of requirements to verify if the system
is suggesting content according to the user profile.
There were created three profiles of users with
different tastes about movies, were also registered a
number of movies and genres in order to crossing
the user's profile with the movies database.
For the content transmission module experiments
were evaluated by the decision variables vLD, vSD,
vHD, vP1, vP2, aST and a51 through the results
obtained by metaheuristics GA and TS.
4.1 Results Obtained by Content
Suggestion Module
Algorithms related to the content suggestion module
could be found in the literature as in Fernandez et al.
(2006), De Paula, Villaça and Magalhães (2011).
The following are presented the steps and the
pseudo-code of the algorithms developed to
authenticate the users and to suggest content using
SPARQL (SPARQL, 2008).
Step 1: The following shows a SPARQL pseudo-
code to authenticate the user. In the case of the
password, we chose to encrypt using MD5.
SELECT ?o
WHERE {
?s w:hasNameValue ?o .
?s w:hasPasswordValue :password .
?s w:hasLoginValue :login
}
Step 2: The following query is run in the user’s
ontology to find out what their preferred genres.
SELECT ?o
WHERE {
?s w:hasPreferedGenre ?o .
?o w:hasGenreNameValue ?n .
?s w:hasLoginValue :login
}
Step 3: Once having the list of user’s preferred
genres, we selected in the ontology server, the films
that are available and they are ordinated by the
viewing rate.
SELECT *
WHERE {
?s w:hasNomeValue ?o .
?s w:hasTaxaVisualizacaoValue ?t .
?s w:hasGenero ?g .
DATA2012-InternationalConferenceonDataTechnologiesandApplications
206
?g w:hasNomeGeneroValue ?n .
filter (?n = :listaGeneros)
}
order by desc(?t)
Step 4: The suggestion is sent to the user. Since it
can accept one of the suggestions, in case of
acceptance the film is transmitted to the user and the
user ontology is updated with the watched film.
INSERT DATA {
w: :user
w:hasWatchedMovie
w: :movie
}
It may be noted that whenever a user accepted a
suggestion, the ontology is updated. Thus, the
system can adapt and evaluate the films that the user
are watching, and so improve the suggestions.
4.2 Results Obtained by Content
Transmission Module
This module runs in an interval of time on the
multimedia application server and contains software
agents that will coordinate the metaheuristics to
optimize the content to be transmitted according to
the number of users connected to the server and
according to the user’s profiles.
The Table 1 shows the first experiment with the
Tabu Search with a small instance of the problem, in
this case are three connected clients of every type
(LD, P1, SD, P2 and HD), a total of 15 clients. From
these clients, 5 receive stereo audio and 10 receive
5.1 audio. The server’s bandwidth was limited to 70
Mbit/s. The value of the objective function (OF) is
70.00 Mbit/s and the harmonic mean is 0.507 Mbit/s.
Table 1: Tabu search results.
Harmonic Nr.Clients
LD 0,710 0,460 0,470 0,525 3
P1 1,140 1,420 1,520 1,340 3
SD 3,610 2,010 3,390 2,805 3
P2 6,130 5,380 5,070 5,492 3
HD 10,030 12,560 10,160 10,801 3
ST 0,180 0,100 0,100 0,110 0,150 0,117 5
A51 0,390 0,500 0,410 0,410 0,490 ‐ 5
A51 0,470 0,390 0,740 0,750 0,750 0,495 5
OF=> 70,000 0,507 15
TabuSearch
Figure 4: Metric comparing Lingo, GA and TS.
Figure 4 shows on a graph the metric value, the
harmonic mean of the results obtained by the Lingo
software (Lingo, 2012), AG and BT of the same
instance of Table 1. It can be observed that the worst
results were obtained by the Lingo. GA and TS
obtained similar results of 0.508 and 0.516 Mbit/s.
The graph of Figure 5 shows the standard
deviation obtained for each metaheuristic. The TS
obtained the best result 0.091 Mbit/s and the GA
obtained 0.107 Mbit/s.
Figure 5: Standard deviation.
The metaheuristics were also tested with 15,000
connected customers: 3,000 of each type (LD, P1,
SD, HD and P2). Of these 15,000, 7,500 received
stereo audio quality and 7,500 received 5.1 audio
quality. Figure 6 shows the harmonic mean of GA
and TS. Both have had very close results with
transmission rates of 0.407 Mbit/s (GA) and 0.406
Mbits/s (TS).
Figure 6: Ontology Structure.
5 CONCLUSIONS
This paper presented an architecture based on
ontologies, software agents and metaheuristics
applied to the multimedia service of the Brazilian
Digital television System divided into two modules.
The suggestion module, that suggests content to the
users, and the content optimization module that has
AnArchitecturebasedonOntologies,AgentsandMetaheuristicsAppliedtotheMultimediaServiceoftheBrazilianDigital
TelevisionSystem
207
the function to optimize the content that will be
transmitted to the user.
Through the implementation and the experiments
that were performed, is observed that it is possible
attend the specified requirements. The experiments
showed that the content suggestion was made
according to the user's profile. The content
optimization module that includes the metaheuristics
GA and TS has obtained good solutions. The
validation showed that the metaheuristics obtained
results of good quality making it possible to obtain a
fair distribution of available bandwidth on the
multimedia application server.
In future works the application of the ontology
for other topics will be extended. Will be included
content that represent music shows, music genres
like jazz, rock, pop, dance, among others. Also it
will be developed a hybrid algorithm to perform the
optimization and the obtained result will be
compared to the current results of the Lingo
software, GA and TS.
REFERENCES
Araújo, R. C., Ricarte, I. L. M., 2010. Ontologias Para
Integração de Metadados da Indústria de Broadcast
Multimídia no Contexto da TV Digital. Revista de
Radiodifusão, n. 4.
Bip, 2001. Boletim de Informação Para Publicitários.
Available at: <http://comercial2.redeglobo.com.br/
biponline/BIP/BIP%20577.pdf>. Acessed in May,
2011.
Carvalho, F. B. S., 2006. Aplicação de Transmissão de
Dados via Rede Elétrica Para o Canal de Retorno em
Televisão Digital, Campina Grande, PB.
Chung, Shu-Hsing, WU, Tai-His, Chang, Chin-Chih.,
2010. An efficient tabu search algorithm to the cell
formation problem with alternative routings and
machine reliability considerations. Computers &
Industrial Engineering, In Press, Corrected Proof,
Available online 31 August 2010, ISSN 0360-8352,
DOI: 10.1016/j.cie.2010.08.016.
De Paula, L., Villaça, R., Magalhães, M., 2011.
Organization of multimedia data for conceptual search
based on ontologies. Journal of the Brazilian
Computer Society. Springer London. Volume 17. Pág.
241-254. Issn: 0104-6500.
Fernández, Y., Arias, J., Nores, M.,; Solla, A., Cabrer, M.,
2006 Avatar: An Improved Solution for Personalized
TV based on Semantic Inference. IEEE Trans. on
Consumer Electronics: Vol. 52(1).
Gendreau, M., Laporte, G., Potvin, J. Y., 2002.
Metaheuristics for the Capacitated VRP. In: TOTH, P.;
VIGO, D. The Vehicle Routing Problem. Philadelphia:
Society for Industrial and Applied Mathematics.
Glover, F., 1986. Future paths for integer programming
and links to artificial intelligence. Computers and
Operations Research, Oxford, UK, v. 13, n. 5, p. 533-
549.
Gonçalvez, J. F., Mauricio, G. C., 2004. An evolutionary
algorithm for manufacturing cell formation.
Computers & Industrial Engineering, Volume 47,
Issues 2-3, November 2004, Pages 247-273, ISSN
0360-8352, DOI: 10.1016/j.cie.2004.07.003.
Gonçalvez, F. E. V., Tiberti, A. J., 2006. A group genetic
algorithm for the machine cell formation problem.
International Journal of Production Economics, v.
102, 1-21.
Holland, J. H., 1975. Adaptation in natural and artificial
systems: an introductory analysis with applications to
biology, control, and artificial intelligence. 1. ed.
Michigan: University of Michigan Press.
Lingo, 2012. Lingo oficial site. Available at:
<http://www.lindo.com>. Accessed in February, 2012.
Link, I. C. D. C., 2011. Um Sistema de Codificação de
Vídeo para TV Digital - SBTVD. Master's Degree
Dissertation Proposal – Interdisciplinary Program of
Graduate Studies in Applied Computing. University of
Vale do Rio dos Sinos, Brazil.
Manhães, M. A. R., Shieh, P J., Lamas, A. C.; Macedo, P.
E. O., 2005. Canal de Interatividade em TV Digital.,
Cad CPqD Tecnologia, páginas 29-36.
Simas, E. P. L., 2007. Utilizando Busca Tabu na
Resolução do Problema de Roteamento de Veículos.
Master's Degree Dissertation – Interdisciplinary
Program of Graduate Studies in Applied Computing.
University of Vale do Rio dos Sinos, Brazil.
Sparql, 2008. SPARQL Query Language for RDF, 2008.
Available at: <http://www.w3.org/TR/rdf-sparql-
query/>. Accessed in February, 2012.
Weissheimer, C. H., 2011. Aplicação de Metaheurísticas
para o modelo Internet Protocol Television. Master's
Degree Dissertation Proposal – Interdisciplinary
Program of Graduate Studies in Applied Computing.
University of Vale do Rio dos Sinos, Brazil.
DATA2012-InternationalConferenceonDataTechnologiesandApplications
208
