CHATBOT INTERNAL MULTI-DIVISI BERBASIS RETRIEVAL-AUGMENTED GENERATION MENGGUNAKAN HYBRID RETRIEVER BM25 DAN APPROXIMATE NEAREST NEIGHBOR

FATCHA, IBKA ANHAR (2026) CHATBOT INTERNAL MULTI-DIVISI BERBASIS RETRIEVAL-AUGMENTED GENERATION MENGGUNAKAN HYBRID RETRIEVER BM25 DAN APPROXIMATE NEAREST NEIGHBOR. S1 thesis, Universitas Mercu Buana Jakarta.

[img]
Preview
Text (HAL COVER)
Cover.pdf

Download (936kB) | Preview
[img] Text (BAB I)
Bab 1.pdf
Restricted to Registered users only

Download (134kB)
[img] Text (BAB II)
Bab 2.pdf
Restricted to Registered users only

Download (234kB)
[img] Text (BAB III)
Bab 3.pdf
Restricted to Registered users only

Download (323kB)
[img] Text (BAB IV)
Bab 4.pdf
Restricted to Registered users only

Download (581kB)
[img] Text (BAB V)
Bab 5.pdf
Restricted to Registered users only

Download (130kB)
[img] Text (DAFTAR PUSTAKA)
Daftar Pustaka.pdf
Restricted to Registered users only

Download (132kB)
[img] Text (LAMPIRAN)
Lampiran.pdf
Restricted to Registered users only

Download (1MB)

Abstract

Companies or organizations with multiple divisions often have internal documents separated by division. Frequently, this information and documentation are not well integrated, leading to delays and difficulties in finding relevant information across divisions. This issue slows down the internal knowledge dissemination process. Therefore, this study designs and develops a Multi-Division Internal Chatbot based on RAG (Retrieval-Augmented Generation) capable of answering user queries according to the internal documents available in each division. The Hybrid Retrieval method (BM25 and ANN) is implemented in the system as a search engine to improve the relevance of responses generated by the LLM. The all-MiniLM-L6- v2 model is used as the embedder to produce vector representations of internal documents and user queries. The application is web-based, with a backend structure managing data and retrieval processes, as well as a user interface for conversational interaction and document management. The system evaluation was conducted using the RAGAS framework to measure the quality of retrieval results and chatbot responses, as well as system latency measurements to assess backend performance. The expected outcome is an initial chatbot design that can enhance efficiency in information retrieval and support more connected document management across divisions. Keywords: Chatbot, Multi-Divisi, RAG, Hybrid Retrieval, LLM Perusahaan ataupun organisasi yang memiliki banyak divisi pastinya memiliki dokumen internal yang terbagi setiap divisi, seringkali informasi dan dokumen internal ini belum terintegrasi dengan baik, sehingga menyebabkan kelambatan dan juga kesulitan dalam mencari informasi yang relevan antar divisi. Permasalahan ini memperlambat proses penyebaran pengetahuan di lingkungan internal. Oleh sebab itu, penelitian ini merancang dan mengembangkan Chatbot Internal Multi-Divisi berbasis RAG (Retrieval-Augmented Generation) yang dapat menjawab pertanyaan pengguna sesuai dengan dokumen internal yang tersedia setiap divisi. Mekanisme Hybrid Retrieval (BM25 dan ANN) diimplementasikan dalam sistem sebagai mesin pencarian, dengan harapan meningkatkan relevansi jawaban yang dihasilkan LLM. Model all-MiniLM-L6-v2 digunakan sebagai embedder untuk menghasilkan representasi vektor dari dokumen internal dan pertanyaan pengguna. Aplikasi yang dibuat berbasis web dengan struktur backend yang mengatur data dan proses pencarian, serta antarmuka pengguna untuk interaksi percakapan dan pengelolaan dokumen. Evaluasi sistem dilakukan menggunakan framework RAGAS untuk mengukur kualitas hasil retrieval dan respons chatbot, serta pengukuran latensi sistem untuk menilai kinerja backend. Hasil yang diharapkan dari penelitian ini adalah rancangan awal chatbot internal yang dapat meningkatkan efisiensi dalam mencari informasi dan mendukung pengelolaan dokumen antar divisi secara lebih terhubung. Kata kunci: Chatbot, Multi-Divisi, RAG, Hybrid Retrieval, LLM

Item Type: Thesis (S1)
NIM/NIDN Creators: 41522010137
Uncontrolled Keywords: Chatbot, Multi-Divisi, RAG, Hybrid Retrieval, LLM
Subjects: 000 Computer Science, Information and General Works/Ilmu Komputer, Informasi, dan Karya Umum > 000. Computer Science, Information and General Works/Ilmu Komputer, Informasi, dan Karya Umum > 004 Data Processing, Computer Science/Pemrosesan Data, Ilmu Komputer, Teknik Informatika
000 Computer Science, Information and General Works/Ilmu Komputer, Informasi, dan Karya Umum > 000. Computer Science, Information and General Works/Ilmu Komputer, Informasi, dan Karya Umum > 005 Computer Programmming, Programs, Data/Pemprograman Komputer, Program, Data > 005.8 Computer Security, Data Security/Keamanan Komputer, Keamanan Data
000 Computer Science, Information and General Works/Ilmu Komputer, Informasi, dan Karya Umum > 020 Library and Information Sciences/Perpustakaan dan Ilmu Informasi > 025 Operations, Archives, Information Centers/Operasional Perpustakaan, Arsip dan Pusat Informasi, Pelayanan dan Pengelolaan Perpustakaan > 025.1 Administration and Library Management/Administrasi dan Manajemen Perpustakaan > 025.11 Finance/Keuangan
Divisions: Fakultas Ilmu Komputer > Informatika
Depositing User: khalimah
Date Deposited: 01 Sep 2026 05:16
Last Modified: 01 Sep 2026 05:16
URI: http://repository.mercubuana.ac.id/id/eprint/103539

Actions (login required)

View Item View Item