TY  - THES
N1  - Pembimbing: Dr. Agung Fatwanto, S.Si, M.Kom
ID  - digilib59064
UR  - https://digilib.uin-suka.ac.id/id/eprint/59064/
A1  - Herwinsyah, NIM.: 21206052001
Y1  - 2023/03/16/
N2  - A good document summary produced by a language model should have a high level of similarity to a summary generated by a human, as the latter considers the semantic value of the text. Typically, a summarization algorithm implemented by a system tends to focus solely on the arrangement of words, without taking into account the semantic aspect of a sentence. The objective of this study is to measure the quality of the summary results obtained from three language models, namely IndoBERT.
In this study, the researcher summarized the same source material using three language models that are focused on the Indonesian language. The summary results generated by the three language models were then compared with a reference summary, which is a summary generated by a human from the same paragraph source. Three evaluation tools were used to measure the quality of the summary results. The study found that the IndoBERT_3 language model, or the IndoBERT version developed by Sarah Lintang (UGM Jogja), produced a summary result that closely approximates the level of similarity to the human-generated summary (gold summary).
PB  - UIN SUNAN KALIJAGA YOGYAKARTA
KW  - Model Bahasa
KW  -  IndoBERT
KW  -  ROUGE
KW  -  BLEU
KW  -  BERTScore
M1  - masters
TI  - PERBANDINGAN KINERJA 3 MODEL BAHASA INDOBERT UNTUK PERINGKASAN TEKS PADA DOKUMEN BAHASA INDONESIA
AV  - restricted
EP  - 128
ER  -