This document describes the design and implementation of a visualization tool to provide more information about the impact of academic publications than citation counts alone. It represents references, citations, and self-citations over time and allows viewing individual papers or collections in different contexts. The tool uses shape grammars to classify papers by impact level into glyphs for compact overviews. It has been implemented in D3 and deployed on the INSPIRE platform for high energy physics publications. Future work may include a Chrome extension and exploring better impact metrics.

1. Visualization of Publication Impact
Eamonn Maguire (CERN), Javier Martin Montull (CERN), Gilles Louppe (CERN & NYU)
CERN, Geneva, Switzerland
New York University, NYC, USA
eamonn.maguire@cern.ch
@antarcticdesign

2. The Problem
Design
Shape Grammars
Implementation
Future Steps
Outline

3. The Problem
Design
Shape Grammars
Implementation
Future Steps
Outline

4. The Problem
When looking through a set
of papers, e.g. when
deciding on who should get a
faculty position, people look
at very coarse metrics, e.g.
the citation count for a paper
and the h-index of an author.
But these metrics don’t tell
the whole story…

5. For each paper, our
system has:
1) every reference with
its citation count; and
2) every citation with its
respective citation
count,
The Problem

6. But the only metric ever really used to evaluate a paper is its
citation count.
But it’s difficult to see the importance of papers citing this paper.
And it’s difficult to see the relative importance of a paper within
its field (consider the importance of references too).

7. The Problem
Representing all of the information just as a number would remove information.
Our goal is to provide more information…visually.

8. The Problem
The visualization needs to encode:
1) The distribution of references and citations over time;
2) The number of citations for everything;
3) Encode self citations; and
4) Allow for extension (e.g. add data).
Encoding Requirements

9. The Problem
Also, we wished to create a design that could be used in a number of
different contexts, including:
1) in a detailed view;
2) as a glyph; and
3) as a summary for an author or research field.
Visualisation Requirements

10. The Problem
Design
Shape Grammars
Implementation
Future Steps
Outline

11. Referenced Papers Cited by
2001
Self
Design the impact graph

12. Design
2001
2001
the impact graph
1) Plot the paper by its date of
publication and citation count.
2) Plot all the references by their
date and citation count.

13. 2001 2001
Design the impact graph
3) Plot all the citing articles by
their date and citation count.
4) Plot the citation momentum
graph. e.g. aggregation of
number of citations in a year.

14. Design
Referenced Papers Cited by
2001
Self
Referenced Papers Cited by Self
Detailed impact graph Impact Glyph Impact Collection

15. 2001
Citations per month/year
Compressed graph structure
Design the impact glaph (graph + glyph = glaph)

16. Design the impact collection
2001
b) Scale Edges
by time
a) Prune Edges
c) Place in context
with other
publications

17. The Problem
Design
Shape Grammars
Implementation
Future Steps
Outline

18. Shape Grammar
Can one quickly see a publications impact just from the shape of the graph?
Low impact High impact
N1 N5
References
Citations
N3 N4N2
Higher Impact
None
References
Citations
Higher Impact
Lower Impact
References
Citations
Average
Average
References
Citations
Lower Impact
Higher Impact
References
Citations
Lower Impact
Lower Impact

19. Low impact High impact
Shape Grammar

20. Low impact High impact
Impact in field
Shape Grammar

21. Shape Grammar
We can use these shapes to reduce the visual complexity of our impact graph collection
visualisation to support higher numbers of papers.
High impact publications

22. The Problem
Design
Shape Grammars
Implementation
Future Steps
Outline

23. Implementation
Created in D3 Installable with
bower or npm
Open source
github.com/inspirehep/impact-graphs

24. Added to the new version on inspire, the largest high energy physics publication resource,
developed largely at CERN and used by over 43,000 physicists last month.
Usage

25. Implementation deployment Core visualisation for INSPIRE

27. The Problem
Design
Shape Grammars
Implementation
Deployment
Future Steps
Outline

28. User Evaluation Chrome Extension
for auto insertion of
glyphs for HEP
papers
Towards a better
impact metric
Future Work

29. Questions?
github.com/inspirehep/impact-graphs
eamonn.maguire@cern.ch
Also later on twitter at @antarcticdesign