VRD: Visual Research Discovery—A Hybrid Framework Integrating Semantic Search and Interactive Visualization

The explosive growth of academic literature is creating major challenges for researchers who are trying to sort through the overwhelming number of publications available. This study proposes an easy-to-use hybrid recommendation system, to help with identifying and accessing relevant research papers....

Full description

Saved in:
Bibliographic Details
Published in:SN computer science Vol. 6; no. 8; p. 958
Main Authors: Yadav, Neha, Gopinathan, Dhanalekshmi
Format: Journal Article
Language:English
Published: Singapore Springer Nature Singapore 01.12.2025
Springer Nature B.V
Subjects:
Applications programs
Bibliographic coupling
Bibliometrics
Citation analysis
Co authorship
Collaboration
Computer Imaging
Computer Science
Computer Systems Organization and Communication Networks
Data Structures and Information Theory
Information overload
Information Systems and Communication Service
Interdisciplinary aspects
Keywords
Knowledge representation
Methods
Natural language processing
Original Research
Pattern Recognition and Graphics
Recommender systems
Researchers
Semantic analysis
Semantics
Software Engineering/Programming and Operating Systems
Vision
Visualization
BERT
Semantic analysis
TF-IDF and cosine similarity
Knowledge graphs
Recommendation system
ISSN:2661-8907, 2662-995X, 2661-8907
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The explosive growth of academic literature is creating major challenges for researchers who are trying to sort through the overwhelming number of publications available. This study proposes an easy-to-use hybrid recommendation system, to help with identifying and accessing relevant research papers. Our system integrated three different approaches - content-based filtering which matches papers based on similar terms, the use of BERT for semantic analysis to encompass more specific structural connections, and knowledge graph methods that show how words relate structurally inside different research communities. The system has been implemented as a simple web application where researchers can enter a title or ID of a research paper, and receive a list of recommended items, with a relevance score. Our framework also suggested an interactive visualization feature to help users explore the relationships of papers, authors, and topics visually. The system’s output was measured using evaluation metrics which support the systems usefulness with a Precision of 0.855, Recall of 0.78, and F1-Score of 0.811 - and a Mean Reciprocal Rank (MRR) value of 0.92, and Normalized Discounted Cumulative Gain (NDCG) value of 0.88 providing confidence that relevant papers would generally be located at the top of the recommended list. When bridging the gap between algorithmically driven recommendations and intuitive exploration, our system can allow researchers to discover relevant literature more efficiently than typical keyword-based search processes.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:2661-8907
2662-995X
2661-8907
DOI:10.1007/s42979-025-04399-y