Skip to content
Switch branches/tags

Latest commit


Git stats


Failed to load latest commit information.
Latest commit message
Commit time

R wrappers for the PLoS Search and ALM API

The Open Access publisher Public Library of Science (PLoS) is providing two APIs for their Search and Article Level Metrics (ALM) API. rplos is a set of R scripts by rOpenSci to facilitate talking to these APIs. The plosOpenR scripts enhance rplos by providing some higher level functions, including visualizations of the results.

Analyzing the article level metrics for a set of PLoS articles involves three steps:

  • Retrieve a set of articles through the Search API
  • Collect the metrics for these articles
  • Visualize the metrics

One incentive for these R wrappers is to further improve rplos by the addition of PLoS funding information. Throughout, this will be showed exemplary by EC funded research visible in the PLoS domain. The FP7 funded project OpenAIRE has the objective to set up a Open Access Infrastructure for Research in Europe. OpenAIRE exposes EC funding information stored in CORDIS via its OAI-PMH Interface. These data can be reused to identify EC funded contributions in the PLoS domain.

Retrieve a set of articles through the PLoS Search API

The articlesSearch script can query the PLoS Search API through a variety of criteria, including:

  • Title
  • Author
  • Editor
  • Affiliation
  • Funder (through the financial disclosure field)

The results can be filtered by date, article type and main subject category.

Additionally, the function plosSearchFinancial provides an alternative interface to query the PLoS Search API by a Funder.

The so retrieved funding information can be matched against data provided by the funder. In the case of EC FP7 publications, projectsFetch.R queries OpenAIRE OAI-PMH interface for EC funded Projects. It returns:

  • Grant ID
  • Project Acronym
  • SC 39 closure

grantFetch.R merges EC and PLoS data tables by the Grant ID.

Collect the metrics for these articles

The almSearch script takes the output of the articlesSearch script as input, but can use any list of PLoS DOIs. The script generates a table of article level metrics for these DOIs and stores them in a CSV file.

Retrieving the metrics for a large set of articles (> 250) can take a long time, and more than 1000 DOIs should probably not retrieved at a time.

The almEventSearch script collects individual events (bookmarks, tweets, etc.) - great for a time series analysis. Only CiteULike and Twitter are currently supported through the API.

The counterFetch.R script collects monthly usage events for a single PLoS article.

Visualize the metrics

Article level metrics are much easier to understand through visualizations. The following visualizations are available:

  • Word cloud
  • Bubble chart
  • Heat map
  • Density plot
  • Time Series Usage events

The scripts that generate these visualizations take the output of the almFetch script. Time Series Usage events is invoked by counterFetch.R script.


Example visualizations are available in the examples folder.


No description, website, or topics provided.



No releases published


No packages published