Path: Top Journal Jurnal_Teknologi_&_Manajemen_Informatika 2007

Penyaringan isi halaman web pornografi dan kekerasan menggunakan algorita kohonen self organizing maps

urnal Teknologi & Manajemen Informatika, Volume 5, Nomor 1, April 2007
Journal from JIPTUNMERPP / 2012-10-01 11:49:58
Oleh : Eddy Sugijono Gonawan ; Arif Djunaidy, Faculty of Information Technology Merdeka University Malang
Dibuat : 2007-04-01, dengan file

Keyword : Web page content filtering, neural network, kohonen self organizing maps, pornography, violence.

Internet has been widely known as one of communication facilities and as information provider that is easier to be accessed by society. Basically, many kind of information can be obtained freely from Internet without any restrictions. Therefore, it makes some people trying to restrict the access of some unexpected information such as pornography any violence. Therefore, the filtering of information is needed if someone or an institution needs to limit the access of some information from internet. This thesis discusses about the research which is related to the development of filtering English web page content system which is predicted consisting information related to pornography and violence. Artificial neural network (ANN) Kohonen Self Organizing Maps (KSOM) is employed to develop the filtering system. A set of pornography and violence keywords were collected from several sources to build the best ANN-KSOM topology. The training process of ANN-KSOM uses the set of keywords that has been previously validated and its weight has been determined. Some parameters of ANN-KSOM topology are tuned up in order to obtain the best ANN-KSOM topology. This ANN-KSOM filtering system is then integrated to an internet browser to provide a web page content filtering system that is ready to apply. The web page content filtering system that has bee successfully developed is currently integrated with an Opera Internet browser. Offline and online tests are performed in order to evaluate its performance. By using 227 pornography and violence keywords that have been validated, the offline test obtains 83% accuracy for 3000 web page samples, while 96.93% accuracy is obtained for the online test using 98 web page samples that are directly accessed through the web browser.

Deskripsi Alternatif :

Internet has been widely known as one of communication facilities and as information provider that is easier to be accessed by society. Basically, many kind of information can be obtained freely from Internet without any restrictions. Therefore, it makes some people trying to restrict the access of some unexpected information such as pornography any violence. Therefore, the filtering of information is needed if someone or an institution needs to limit the access of some information from internet. This thesis discusses about the research which is related to the development of filtering English web page content system which is predicted consisting information related to pornography and violence. Artificial neural network (ANN) Kohonen Self Organizing Maps (KSOM) is employed to develop the filtering system. A set of pornography and violence keywords were collected from several sources to build the best ANN-KSOM topology. The training process of ANN-KSOM uses the set of keywords that has been previously validated and its weight has been determined. Some parameters of ANN-KSOM topology are tuned up in order to obtain the best ANN-KSOM topology. This ANN-KSOM filtering system is then integrated to an internet browser to provide a web page content filtering system that is ready to apply. The web page content filtering system that has bee successfully developed is currently integrated with an Opera Internet browser. Offline and online tests are performed in order to evaluate its performance. By using 227 pornography and violence keywords that have been validated, the offline test obtains 83% accuracy for 3000 web page samples, while 96.93% accuracy is obtained for the online test using 98 web page samples that are directly accessed through the web browser.

Beri Komentar ?#(0) | Bookmark

PropertiNilai Properti
ID PublisherJIPTUNMERPP
OrganisasiF
Nama KontakDra. Wiwik Supriyanti, SS
AlamatJl. Terusan Halimun 11 B
KotaMalang
DaerahJawa Timur
NegaraIndonesia
Telepon0341-563504
Fax0341-563504
E-mail Administratorperpus@unmer.ac.id
E-mail CKOwsupriyanti@yahoo.com

Print ...

Kontributor...

  • Editor: Wiwik Supriyanti, Dra. SS.