Implementacija RAG sistema za iskanje informacij v nestrukturiranih dokumentih z uporabo velikih jezikovnih modelov : diplomsko delo

Jovanović, Nik

| | SLO | ENG | Cookies and privacy

Bigger font | Smaller font

First page > Show document

Show document

Title:	Implementacija RAG sistema za iskanje informacij v nestrukturiranih dokumentih z uporabo velikih jezikovnih modelov : diplomsko delo
Authors:	ID Jovanović, Nik (Author) ID Šumak, Boštjan (Mentor) More about this mentor...
Files:	VS_Jovanovic_Nik_2025.pdf (2,37 MB) MD5: 8C00B44B53E4E7C4292D2E1E2D18713E
Language:	Slovenian
Work type:	Bachelor thesis/paper
Typology:	2.11 - Undergraduate Thesis
Organization:	FERI - Faculty of Electrical Engineering and Computer Science
Abstract:	V zaključnem delu obravnavamo problem iskanja informacij iz nestrukturiranih dokumentov, kjer osnovni veliki jezikovni modeli pogosto vračajo izmišljene oziroma netočne odgovore. Kot rešitev smo uporabili arhitekturo generacije z iskanjem v zunanjih virih (Retrieval-Augmented Generation – RAG). Cilj naloge je bil zasnovati in implementirati sistem, ki zmanjša pojav halucinacij pri poizvedbah. Sistem smo testirali na desetih računih v obliki PDF dokumentov, pri čemer smo uporabili pet skrbno pripravljenih vprašanj. Rezultati kažejo zmanjšano pojavnost halucinacij, a hkrati opozarjajo na omejitve, saj bi bilo za zanesljivejše zaključke potrebnih več dokumentov in obsežnejše testiranje.
Keywords:	RAG, umetna inteligenca, vektorske baze, react, next.js
Place of publishing:	Maribor
Place of performance:	Maribor
Publisher:	[N. Jovanović]
Year of publishing:	2025
Number of pages:	1 spletni vir (1 datoteka PDF (X, 57 str.))
PID:	20.500.12556/DKUM-94742
UDC:	004.832.2(043.2)
COBISS.SI-ID:	260364291
Publication date in DKUM:	23.09.2025
Views:	158
Downloads:	41
Metadata:
Categories:	KTFMB - FERI
:	Copy citation

Average score:	(0 votes)
Your score:	Voting is allowed only for logged in users.
Share:

Hover the mouse pointer over a document title to show the abstract or click on the title to get all document metadata.

Licences

License:	CC BY-NC-ND 4.0, Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International

Link:	http://creativecommons.org/licenses/by-nc-nd/4.0/
Description:	The most restrictive Creative Commons license. This only allows people to download and share the work for no commercial gain and for no other purposes.
Licensing start date:	26.08.2025

Secondary language

Language:	English
Title:	Implementation of a RAG system for information retrieval in unstructured documents using large language models
Abstract:	This thesis addresses the problem of retrieving information from unstructured documents, where basic large language models often return fabricated or inaccurate answers. As a solution, we applied the Retrieval-Augmented Generation (RAG) architecture, which integrates external knowledge sources into the response generation process. The main goal was to design and implement a system that reduces the occurrence of hallucinations in queries. The system was tested on ten PDF invoices using five carefully prepared questions. The results indicate a reduced occurrence of hallucinations but also highlight limitations, as more documents and extensive testing would be required for reliable conclusions.
Keywords:	RAG, artificial intelligence, vector database, react, next.js

Comments

Leave comment

You must log in to leave a comment.

Comments (0)

0 - 0 / 0

There are no comments!

Back