Verdianto, Satrio (2016) Strategi Pemilihan Kalimat Pada Peringkasan Multi Dokumen Satrio Verdianto Nrp 5111 100 183. Undergraduate thesis, Institut Teknologi Sepuluh Nopember Surabaya.
Preview |
Text
5111100183-Undergraduate_Thesis.pdf - Published Version Download (1MB) | Preview |
Abstract
Ringkasan berita diartikan sebagai teks yang dihasilkan dari satu atau lebih kalimat yang menyampaikan informasi penting dari berita. Salah satu fase penting dalam peringkasan adalah pembobotan kalimat (sentence scoring). Dimana pada peringkasan berita, metode pembobotannya sebagian besar menggunakan fitur dari berita sendiri. Berdasarkan hasil dari penelitian (Ferreira, et al., 2014) bahwa untuk pembobotan kalimat pada dokumen yang memiliki karakter teks pendek dan terstruktur seperti berita maka teknik pembobotan kalimat terbaik adalah dengan menggunakan kombinasi dari keempat fitur yaitu word frequency, TF-IDF, posisi kalimat, dan kemiripan kalimat terhadap judul (Resemblance to the title ).
Pada penelitian ini kombinasi keempat fitur tersebut dibandingkan dengan kombinasi tiga fitur dan dua fitur dan dievaluasi menggunakan nilai ROUGE-N dan dievaluasi berdasarkan lama waktu eksekusi. Berdasarkan hasil uji coba didapatkan hasil bahwa yang paling optimal diantara keempat kombinasi fitur tersebut adalah kombinasi antara dua buah fitur yakni fitur posisi kalimat dan word frequency dengan nilai ROUGE-N sebesar 0.679 dan lama waktu eksekusi 28.458 detik
=============================================================================================
Summary of news is defined as a text resulting from one or more sentences that convey important information from news. One important phase in text summarization is weighting sentence (sentence scoring). In the news summarization the weighting method mostly using the features of the news itself. Based on the results of the study (Ferreira, et al., 2014) that for weighting sentences in documents that have character short text and structured as news, the technique of weighting sentence is best to use a combination of all four features that word frequency, TF-IDF, position, and Resemblance to the title.
In this study, the combination of four features compared to the combination of three features and two features and evaluated using a value ROUGE-N and evaluated based on the length of time of execution. Based on test results showed that among the four combination of these feature, the most optimal combination is the combination of two features those are position of the sentence feature and word frequency feature with ROUGE-N 0.679 and length of time of execution 28.458 sec
| Item Type: | Thesis (Undergraduate) |
|---|---|
| Additional Information: | RSIf 005.133 Ver s |
| Uncontrolled Keywords: | kemiripan kalimat terhadap judul, pembobotan kalimat, posisi kalimat, ROUGE-N, TF-IDF, word frequency |
| Subjects: | T Technology > T Technology (General) > T57.5 Data Processing |
| Divisions: | Faculty of Information Technology > Informatics Engineering > 55201-(S1) Undergraduate Thesis |
| Depositing User: | EKO BUDI RAHARJO |
| Date Deposited: | 06 Jul 2020 06:12 |
| Last Modified: | 06 Jul 2020 06:12 |
| URI: | http://repository.its.ac.id/id/eprint/76308 |
Actions (login required)
 |
View Item |