Visualizing Article Level Metrics in d3.js

Presentation given at the ALM Workshop in San Francisco, October 10-12, 2013

When scouring the literature for hidden gems, readers must be able to quickly dissern the value of a research product. Skilled readers rely on a series of cues from the product page in order to pass judgement, even before engaging with the text at hand. In more recent years, with the advent of Article Level Metrics (ALMs), readers started ahving a new set of value signals at their disposal. ALMs provide metrics of peer use, ranging from downloads to citations to social media. Yet, while ALMs and altmetrics have been growing in popularity, there has been little attention on how these metrics should be displayed as to maximize their utility for readers.

Inspired by the "sparkline" notion that data should be displayed in a way that is succinct and design-simple, the "almviz" group at the altmetrics hackaton (2012) set out some guiding principles for visualizing ALMs so that they would be most helpful to readers. These principles were used for an implementation using data from PLOS' Article Level Metrics application and the javascript visualization library d3.js. The implementation deemphasizes absolute numbers in favour of helping readers identify variation over time, under the belief that raw counts are less meaningful to a reader than information on its pattern of use. This presentation will showcase the simple, yet carefully designed visualization, along with the justifications for the design decisions.

As readers are exposed to a consistent visualization accross journals using the Public Knowledge Project's "Open Journal Systems", it is expected that they will quickly learn to glean peer's usage across various dimensions with ease. The search for those prized research gems has just been made a little easier.

Transcript

1. VISUALIZING ALM’S IN D3.JS Juan Pablo Alperin (@juancommander)
2. THE CHALLENGE to improve the metrics displayed on publisher pages NOT for data analysis or research NOT for aggregate metrics rather for helping the reader understand the article’s impact
3. THE COMPETITION not very stiff it appears as if little thought has gone into displaying metrics maybe unfair? focus has been on displaying numbers
4. I AM NO PICASSO =
5. I AM NO MIKE BOSTOCK = creator of d3.js
6. EXHIBIT I – PLOS
7. EXHIBIT I – PLOS
8. EXHIBIT II – ALTMETRIC.COM http://www.flickr.com/photos/yamagatacamille/5895167009/
9. EXHIBIT II – ALTMETRIC.COM
10. EXHIBIT II – ALTMETRIC.COM
11. EXHIBIT III – IMPACTSTORY
12. EXHIBIT III – IMPACTSTORY
13. EXHIBIT IV – ELIFESCIENCES.ORG
14. EXHIBIT IV – ELIFESCIENCES.ORG
15. OK. ENOUGH CRITIQUE.
16. MONEY WHERE MOUTH IS http://www.flickr.com/photos/yamagatacamille/5895167009/
17. LET’S SEE A LIVE DEMO also on any of the co-action journals
18. THE NUMBER design considerations how many!
19. THE SPARKLINE design considerations sense of history
20. A SENSE OF SCALE design considerations maximum value
21. MULTIPLE AGGREGATIONS design considerations all levels available
22. MINIMALIST design considerations •few labels •no grid lines
23. MORE INFORMATION design considerations link to source data tooltip
24. STACKED design considerations
25. OTHER CONSIDERATIONS in current version • turn off graphics for few events future versions • add graphics based on when service queried if no timestamp is available • allow showing/hiding of sources • link through to more complete article profile
26. CREDITS almviz hackathon group (ALM workshop 2012) but especially Martin Fenner