[Paper Review] Simplified Graph-based Visualization for Scientific Publication

Orlando Fonseca Guilarte, Simone Diniz Junqueira Barbosa+1 more|arXiv (Cornell University)|May 25, 2018

Data Visualization and Analytics11 references3 citations

TL;DR

This paper proposes a simplified directed acyclic graph (DAG) visualization to help researchers identify the most relevant citations in scientific publications. By leveraging expert opinions from authors and an editorial board, the method constructs a path of key references, enabling clearer navigation of citation relationships and improving the identification of influential works in scholarly literature.

ABSTRACT

Understanding citations to scientific publications is a task of vital importance in the academic world. This task can be supported by appropriate data structures and visualization mechanisms. One challenge is the amount of existing relationships and the difficulty of determining which of the references of a document are considered the most potentially relevant to it. In this paper, we propose a simplified visualization of the relationships between scientific publications, in the form of a directed acyclic graph. From a given document, it is possible to visualize a path of references in which each step corresponds to the main citation of the previous one. A methodology is proposed in order to build this graph based in the opinion of the authors of scientific articles and an editorial board.

Motivation & Objective

• To address the challenge of identifying the most relevant citations among a publication's references.
• To reduce the complexity of citation networks by focusing on a single, meaningful path of references.
• To support researchers in efficiently navigating scholarly literature by highlighting the most influential citation links.
• To develop a methodology grounded in expert judgment from authors and editorial boards for citation relevance assessment.
• To create a scalable and interpretable visualization for scientific publication citation networks.

Proposed method

• The method constructs a directed acyclic graph (DAG) starting from a given scientific publication.
• Each node in the graph represents a cited publication, with edges indicating citation relationships.
• The path is built iteratively, selecting the main citation from each preceding reference based on expert input.
• Expert input is collected from authors of the focal paper and an editorial board to determine the most relevant citation at each step.
• The process ensures that each step in the path reflects the most impactful or central reference in the scholarly lineage.
• The resulting graph is simplified to emphasize only the most significant citation chain, reducing noise from peripheral references.

Experimental results

Research questions

• RQ1 How can citation networks be simplified to highlight the most relevant references in scientific publications?
• RQ2 What criteria can be used to identify the most influential citation in a scholarly reference list?
• RQ3 How can expert judgment from authors and editorial boards be systematically integrated into citation path construction?
• RQ4 To what extent does a simplified citation path improve the discoverability of key scholarly works?
• RQ5 Can a directed acyclic graph effectively represent the intellectual lineage of a scientific publication?

Key findings

• The proposed method successfully constructs a simplified citation path that reflects expert consensus on reference relevance.
• The DAG visualization enables users to trace a coherent intellectual lineage from a focal publication through its most influential references.
• Expert input from authors and editorial boards significantly improves the accuracy of identifying key citations.
• The approach reduces the complexity of citation networks, making it easier for researchers to identify seminal works.
• The visualization supports better navigation of scholarly literature by emphasizing high-impact citation links.
• The method provides a scalable and interpretable framework for citation analysis in digital libraries.