Mustaqhfiri, Muchammad (2011) Peringkasan Teks Otomatis Berita Olahraga Berbahasa Indonesia menggunakan Metode Maximum Marginal Relevance. Undergraduate thesis, Universitas Islam Negeri Maulana Malik Ibrahim.
Text (Fulltext)
06550057.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (5MB) | Request a copy |
Abstract
ABSTRAK
Seiring perkembangan teknologi informasi mengakibatkan teknologi internet semakin pesat, sehingga banyak berita online khususnya berita olahraga. Dengan adanya peringkasan teks secara otomatis ini diharapkan membantu mengurangi waktu membaca keseluruhan isi berita dengan hanya membaca hasil ringkasannya, sehingga memudahkan dalam mencari informasi berita olahraga.
Penelitian ini diawali dengan proses text preprocessing, yaitu pemrosesan teks untuk mendapatkan term kata. Metode ini terdiri dari case folding, pemecahan kalimat, filtering, tokenizing, dan stemming. Hasil dari proses ini kemudian dihitung bobot tf-idf, bobot relevance dan bobot similarity. Untuk menghasilkan ringkasan dilakukan proses ekstraksi yaitu menghitung bobot maximum marginal relevance kalimat dari kombinasi cosine similarity, yaitu relevance dan similarity. Metode ekstraksi maximum marginal relevance merupakan metode yang digunakan dalam mengurangi redudansi kalimat dalam dokumen dalam menentukan sebagai ringkasan.
Data uji coba untuk pengujian akan diambilkan dari surat kabar berbahasa Indonesia on-line. Dari hasil pengujian kemudian dievaluasi dengan hasil ringkasan manual dan ringkasan sistem peringkasan otomatis lain. Hasil dari evaluasi dengan ringkasan manual menghasilkan rata-rata recall 60%, precision 77%, dan f-measure 66%. Sedangkan hasil evaluasi dengan sistem peringkasan otomatis lain menghasilkan rata-rata recall 79%, precision 89% dan f-measure 82%.
ABSTRACT
Along with the development of information technologies resulted in internet technology more rapidly, so a lot of online news, especially sports news. With the automatic text summarization is expected to help reduce the time to read this entire story by just reading the summary, making it easier to find information in sports news.
This study begins with the text preprocessing, ie processing of text to get the word term. This method consists of case folding, split sentences, filtering, tokenizing and stemming. The results of this process is then calculated weighted tf-idf, weighting relevance and weight of similarity. To produce a summary of the extraction process is performed to calculate the maximum weight of a combination of marginal relevance sentence cosine similarity, ie relevance and similarity. Methods for extracting maximum marginal relevance is the method used in reducing redundancies in determining the sentence in the document as a summary.
Trial data for testing will be deducted from the Indonesian-language newspaper on-line. From the test results then evaluated with a summary of the results of manual and automated summary peringkasan other systems. The results of the evaluation with a summary of the average man produces 60% recall, precision 77% and 66% f-measure. While the evaluation results with other automated systems peringkasan yield an average recall of 79%, precision 89% and 82% f-measure.
Item Type: | Thesis (Undergraduate) | |||||||||
---|---|---|---|---|---|---|---|---|---|---|
Supervisor: | Abidin, Zainal and Kusumawati, Ririen | |||||||||
Contributors: |
|
|||||||||
Keywords: | Peringkasan; Berita; Text Preprocessing; TF-IDF; Relevance; Similarity; Maximum Marginal Relevance Summarization; News; Text Preprocessing; TF-IDF; Relevance; Similarity; Maximum Marginal Relevance | |||||||||
Departement: | Fakultas Sains dan Teknologi > Jurusan Teknik Informatika | |||||||||
Depositing User: | Moch. Nanda Indra Lexmana | |||||||||
Date Deposited: | 16 May 2023 13:25 | |||||||||
Last Modified: | 16 May 2023 13:25 | |||||||||
URI: | http://etheses.uin-malang.ac.id/id/eprint/50144 |
Downloads
Downloads per month over past year
Actions (login required)
View Item |