AcademicRAG: Knowledge graph enhanced retrieval-augmented generation for academic resource discovery

Thumbnail Image

Files

master_Liu_Zhuchenyang_2025.pdf (6.15 MB)

URL

Journal Title

Journal ISSN

Volume Title

School of Science | Master's thesis
Unless otherwise stated, all rights belong to the author. You may download, display and print this publication for Your own personal use. Commercial use is prohibited.

Authors

Date

2025-05-26

Department

Major/Subject

Data Science

Mcode

Degree programme

Master's Programme in ICT Innovation

Language

en

Pages

79

Series

Abstract

This thesis presents AcademicRAG, a novel knowledge graph-enhanced framework for academic resource discovery that addresses critical limitations in traditional information retrieval systems. By integrating graph-based retrieval mechanisms with large language models, the framework significantly improves the capture, representation, and utilization of complex semantic relationships inherent in academic knowledge. The research introduces two key technical innovations: clue-guided keyword generation and subgraph-based retrieval, which together enhance query performance across diverse academic domains. Comprehensive evaluations demonstrate that AcademicRAG consistently outperforms existing state-of-the-art approaches in response comprehensiveness, diversity, and user empowerment. The practical implementation of a Course Discovery System validates the framework's utility in educational contexts, effectively addressing distinct information needs for both students and faculty. For students, the system provides personalized learning pathways and prerequisite analyses; for faculty, it offers curriculum optimization insights by identifying content overlaps and progression patterns across departmental boundaries. Despite notable achievements, AcademicRAG and its application acknowledges limitations in entity extraction consistency and relationship standardization, suggesting future work in hybrid extraction approaches and multimodal content integration. AcademicRAG represents a significant advancement toward more intuitive, context-aware academic information systems, enabling richer exploration of educational resources and more informed academic decision-making.

Description

Supervisor

Holme, Petter

Thesis advisor

Zea, Elias
Tavallaey, Shiva Sander

Keywords

knowledge graph, retrieval-augmented generation, academic resource discovery, large language models, graph retrieval, course discovery system

Other note

Citation

Permanent link to this item

https://urn.fi/URN:NBN:fi:aalto-202506164710

Collections

[dipl] Perustieteiden korkeakoulu / SCI