Analisis Komparatif Word Embedding LLaMA dan IndoBERT Pada Chatbot Hukum Pidana Berbasis Retrieval Augmented Generation

Triyudanto, Bagus (2026) Analisis Komparatif Word Embedding LLaMA dan IndoBERT Pada Chatbot Hukum Pidana Berbasis Retrieval Augmented Generation. Bachelor thesis, Institut Teknologi Kalimantan.

[img] Text
11211021_cover.pdf
Restricted to Repository staff only until 4 October 2028.

Download (482kB) | Request a copy
[img] Text
11211021_statement_of_authenticity.pdf
Restricted to Repository staff only until 4 October 2028.

Download (340kB) | Request a copy
[img] Text
11211021_publishing_agreement.pdf
Restricted to Repository staff only until 4 October 2028.

Download (404kB) | Request a copy
[img] Text
11211021_aprroval_sheet.pdf
Restricted to Repository staff only until 4 October 2028.

Download (293kB) | Request a copy
[img] Text
11211021_preface.pdf
Restricted to Repository staff only until 4 October 2028.

Download (392kB) | Request a copy
[img] Text
11211021_abstract_id.pdf
Restricted to Repository staff only until 4 October 2028.

Download (367kB) | Request a copy
[img] Text
11211021_abstract_en.pdf
Restricted to Repository staff only until 4 October 2028.

Download (369kB) | Request a copy
[img] Text
11211021_table_of_content.pdf
Restricted to Repository staff only until 4 October 2028.

Download (329kB) | Request a copy
[img] Text
11211021_illustrations.pdf
Restricted to Repository staff only until 4 October 2028.

Download (273kB) | Request a copy
[img] Text
11211021_tables.pdf
Restricted to Repository staff only until 4 October 2028.

Download (326kB) | Request a copy
[img] Text
11211021_chapter_1.pdf
Restricted to Repository staff only until 4 October 2028.

Download (476kB) | Request a copy
[img] Text
11211021_chapter_2.pdf
Restricted to Repository staff only until 4 October 2028.

Download (996kB) | Request a copy
[img] Text
11211021_chapter_3.pdf
Restricted to Repository staff only until 4 October 2028.

Download (723kB) | Request a copy
[img] Text
11211021_chapter_4.pdf
Restricted to Repository staff only until 4 October 2028.

Download (859kB) | Request a copy
[img] Text
11211021_chapter_5.pdf
Restricted to Repository staff only until 4 October 2028.

Download (398kB) | Request a copy
[img] Text
11211021_bibliography.pdf
Restricted to Repository staff only until 4 October 2028.

Download (461kB) | Request a copy
[img] Text
11211021_enclosure.pdf
Restricted to Repository staff only until 4 October 2028.

Download (274kB) | Request a copy
[img] Text
11211021_paper.pdf
Restricted to Repository staff only until 4 October 2028.

Download (457kB) | Request a copy
[img] Text
11211021_presentation.pdf
Restricted to Repository staff only until 4 October 2028.

Download (1MB) | Request a copy

Abstract

Chatbot berbasis kecerdasan buatan (AI) berpotensi merevolusi akses informasi hukum pidana dengan memberikan respons cepat dan akurat melalui pemrosesan bahasa alami (NLP). Dalam konteks hukum Indonesia, kompleksitas terminologi dan ketidakpastian hukum seringkali menjadi hambatan dalam pemahaman masyarakat terhadap peraturan pidana. Meskipun teknologi chatbot telah berkembang dari sistem berbasis aturan menuju model neural networks dan deep learning, tantangan utama tetap pada akurasi respons dalam konteks hukum spesifik. Model generatif umum seperti GPT-3 berisiko memberikan rekomendasi keliru akibat kurangnya pemahaman terhadap terminologi hukum Indonesia. Penelitian ini menunjukkan bahwa pendekatan Retrieval Augmented Generation (RAG) dengan kombinasi model LLaMA dan IndoBERT dapat meningkatkan akurasi respons chatbot hukum pidana. LLaMA dioptimalkan untuk memahami konteks panjang, sementara IndoBERT dirancang untuk menangani terminologi hukum Indonesia secara lebih akurat. Hasil penelitian mengungkapkan bahwa model ini mampu mengurangi risiko 'halusinasi hukum' dalam chatbot dengan memberikan respons yang berbasis pada referensi hukum terpercaya. Pendekatan RAG yang diterapkan juga memungkinkan chatbot untuk secara eksplisit merujuk KUHP 2023 dan yurisprudensi Indonesia, sehingga meningkatkan keandalan informasi yang diberikan kepada pengguna.

Item Type: Thesis (Bachelor)
Subjects: K Law > KZ Law of Nations
Q Science > QA Mathematics > QA75 Electronic computers. Computer science
Divisions: Jurusan Matematika dan Teknologi Informasi > Informatika
Depositing User: Bagus Triyudanto
Date Deposited: 16 Jul 2026 02:49
Last Modified: 16 Jul 2026 02:49
URI: http://repository.itk.ac.id/id/eprint/27427

Actions (login required)

View Item View Item