One of the promises of altmetrics — an approach to measuring attention on research papers that relies on alternative measures to citations, such as downloads, social media mentions and collections in online libraries — is that it could provide an almost real-time view of the papers provoking most excitement. Citations, by contrast, are inevitably slow to gather pace.
So as Nature was starting to think about its review of 2012 (at the end of November — we start early!) we also asked altmetrics experts to pick out the most noted papers of the year. The results show some of the promise — and also the teething problems — of the new kids on the block.
The wizards at Altmetric.com picked out for us the top research articles mentioned directly online. (The company is a product of Digital Science, a sister company to Nature Publishing Group, which publishes Nature.com.) The technology tracks public, direct mentions of a research paper online by picking up DOI references, so it’s important to emphasize that we could not see articles that may have made a huge media splash, but that people did not link to directly — nor the many non-public Facebook links. Also, we knew that articles published earlier in the year would have had time to pick up more mentions — but decided to accept that bias. And here are the top ten, as of 27 November:
- The biological impacts of the Fukushima nuclear accident on the pale grass blue butterfly (2,142 people tweeting; 9 on blogs; 3 reddit posts; 131 Facebook posts and 9 G+ direct DOI links).
- Association of coffee drinking with total and cause-specific mortality (1,620; 4; 11; 40; 18)
- Rape-related pregnancy: estimates and descriptive characteristics from a national sample of women (a 1996 paper) (1,594; 6; 1; 90; 2)
- Food for thought. What you eat depends on your sex and eating companions (a 2009 paper) (1,606; 0; 0; 0; 0)
- Bright minds and dark attitudes: lower cognitive ability predicts greater prejudice through right-wing ideology and low intergroup contact
(1,411; 11; 4; 38; 25)
- Unilateral dermatoheliosis (1,124; 6; 1; 102; 57)
- Higher social class predicts increased unethical behavior (1,218; 3; 9; 7; 12)
- Science faculty’s subtle gender biases favor male students (1,029; 8; 21; 89; 32)
- Measuring the evolution of contemporary western popular music (908; 2; 5; 98; 23)
- Classic Nintendo games are (NP-)hard (904; 9; 4; 10; 62).
For another altmetrics view on 2012, we turned to Mendeley.com, a reference manager and social academic network that counts 2 million users. This time we saw a lot of perspectives and reviews in the list. The most-read papers in 2012 by Mendeley’s readers — defined as those that Mendeley users have added to their document libraries — were not actually published in 2012 (with Uri Alon’s 2009 paper ‘How to choose a good scientific problem‘ at the top of the list, with 5,947 library additions, or ‘readers’, this year). But, asking for only papers published in 2012, the most read as of 30 November were:
- An integrated encyclopedia of DNA elements in the human genome (730 readers)
- A whole-cell computational model predicts phenotype from genotype (569)
- Differential gene and transcript expression analysis of RNA-seq experiments with TopHat and Cufflinks [protocols paper] (492)
- Is cancer a metabolic rebellion against host aging? In the quest for immortality, tumor cells try to save themselves by boosting mitochondrial metabolism [perspective] (477)
- Approaching a state shift in Earth’s biosphere [review] (426)
- Optogenetic stimulation of a hippocampal engram activates fear memory recall (410)
- Biodiversity loss and its impact on humanity [review] (389)
- Landscape of transcription in human cells (377)
- The origin of extracellular fields and currents — EEG, ECoG, LFP and spikes [review] (333)
- Architecture of the human regulatory network derived from ENCODE data (330)
What a difference — with three papers on the list relating to ENCODE. Company co-founder and chief executive Victor Henning says that several studies have examined and confirmed a strong correlation between Mendeley readership statistics and citation figures on services such as Web of Science, Scopus and Google Scholar.
And what did citations tell us? Thomson Reuters (who provided the data for our giant infographic of papers published by the top countries in 2012), had a look at citations amassed by 2012 papers. (In the infographic, the analysts correct for a paper’s age and field when examining citations, but here we look at overall citations amassed by any paper since the start of the year.) At the time of search (late November), only indexed data up to August was available — so this is best thought of as the top-cited papers of the year’s first half (and likely to be among the top even by the end of the year). Once again, we accepted the bias that papers published at the beginning of the year had more time to pick up mentions.
- Combined search for the Standard Model Higgs boson using up to 4.9 fb(-1) of pp collision data at root s=7 TeV with the ATLAS detector at the LHC (140 citations).
- Combined results of searches for the standard model Higgs boson in pp collisions at root s=7 TeV (128)
- Observation of electron-antineutrino disappearance at Daya Bay (104)
- Intratumor heterogeneity and branched evolution revealed by multiregion sequencing (94)
- The Pfam protein families database (90)
- Solution-processed small-molecule solar cells with 6.7% efficiency (81)
- The Atmospheric Imaging Assembly (AIA) on the Solar Dynamics Observatory (SDO) (71)
- Observation of reactor electron antineutrinos disappearance in the RENO experiment (66)
- Global malaria mortality between 1980 and 2010: a systematic analysis (67)
- Tandem polymer solar cells featuring a spectrally matched low-bandgap polymer (64)
Let’s be frank: this is a bit of a mess. Not a single common paper among the 30. All of these papers were at the peak of popularity or citation-strength under different measures — but this exercise does hint at how altmetrics and citations are picking up different facets of attention. And, just perhaps, that you can prove anything you want with statistics. Readers, your comments on how to improve this exercise would be much appreciated.