Implementasi Peringkasan Multidokumen Berita Berbahasa Indonesia Dengan Pemilihan Kata Kunci Twitter Menggunakan Autocorrelation Wavelet Coefficients

Gusman, Oshi Prahtiwi (2016) Implementasi Peringkasan Multidokumen Berita Berbahasa Indonesia Dengan Pemilihan Kata Kunci Twitter Menggunakan Autocorrelation Wavelet Coefficients. Undergraduate thesis, Institut Teknologi Sepuluh Nopember Surabaya.

[img] Text
5112100195-undergraduate-theses-.pdf - Published Version

Download (1MB)

Abstract

Twitter digunakan untuk menyampaikan informasi berupa tweet yang merepresentasikan suatu kejadian yang mengakibatkan munculnya issue. Issue yang paling sering dibahas disebut dengan Trending Issue. Dalam tugas akhir ini, digunakan data twitter yang telah diambil selama bulan April dan Mei 2016 Metode Autocorrelation Wavelet coefficients berguna untuk mendapatkan kata kunci yang muncul secara periodik atau berulang (trivial) yang merepresentasikan kejadian biasa dan akan dieliminasi sehingga menyisakan kata kunci penting (non-trivial). Kata kunci penting digunakan untuk peringkasan berita menggunakan metode pembobotan kalimat berdasarkan trending issue sehingga menghasilkan ringkasan berita yang koheren. Setelah dilakukan pengujian ada beberapa faktor utama yang mempengaruhi hasil ringkasan berita, diantaranya penggunaan keyword yang spesifik, jangkauan lokasi pengambilan tweet, penentuan confidence boundary, dan perlu atau tidaknya proses eliminasi kata kunci trival. Nilai silhouette terbaik ditunjukkan pada hasil ekstraksi trending issue dengan pembuangan kata kunci trivial sebesar 0,36322. Nilai rouge terbaik ditunjukkan pada hasil ringkasan tanpa pembuangan kata kunci sebesar 0,30199. =========================================================== Twitter is used to deliver the information in the form of tweets that represents an event that resulted an issue. The most frequently discussed issue is called Trending Issue. In this Final Project, the twitter data is collected during April and May 2016. Autocorrelation wavelet coefficients method is used to get the keywords that appear periodically. The repeated keywords (trivial) represent a regular event and will be eliminated thus leaving important keywords (non-trivial). Important keywords will be used to summarize news and become the purpose of this Final Project, using the phrase weighting method based trending issue to produce a coherent summary of the news. After testing, a number of key factors that influence the outcome of a news summary, including the use of a specific keyword, scope of the location of the tweet, determination of the confidence boundary, and whether or not the elimination process trival keywords. Best silhouette value shown in the results of extraction trending issue with the disposal amounting to 0.36322 trivial keywords. Best rouge value shown in the summary results without disposal keywords by 0.30199.

Item Type: Thesis (Undergraduate)
Additional Information: RSIf 006.312 Gus i-1
Uncontrolled Keywords: Twitter, Wavelet Coefficients, Autocorrelations, Trending Issue
Subjects: Q Science > QA Mathematics > QA76.76.A65 Application software
Divisions: Faculty of Information and Communication Technology > Informatics > (S1) Undergraduate Theses
Depositing User: EKO BUDI RAHARJO
Date Deposited: 27 Nov 2019 09:04
Last Modified: 27 Nov 2019 09:04
URI: http://repository.its.ac.id/id/eprint/72096

Actions (login required)

View Item View Item