Metrics for scientific software

Collecting metrics is a key issue in terms of scientific software now a days. One of the challenges of collecting metrics for scholarly outputs is persistent identifiers. For journal articles, the Digital Object Identifier (DOI) has become the de-facto standard, other popular identifiers are the arxiv ID for ArXiV preprints, the pmid from PubMed, the identifiers used by Scopus and Web of Science.
For other research outputs the scenario is still less clear. DOIs are also used for datasets, as well as many other identifiers, particularly, in the life sciences.

While collecting metrics for research outputs, the pre-requisites are slightly different. We need identifiers recognized by the services collecting the metrics, not by the data repository or other service that contain the research output. For many services such as social media such as Facebook, Twitter or Reddit, URL is the primary identifier for a resource. This means that we should have one or more URLs for every research output if we want to track the metrics – typically the publisher or data repository landing page. Since URLs can be tricky, Google, Facebook, Twitter and others have come up with the concept of a Canonical URL. However, some care should be taken into constructing a proper canonical URL.

In general, URLs are good enough to start gathering metrics for scholarly outputs. Scientific software is a good example where persistent identifiers are not commonly used, but we can still collect many meaningful metrics using the repository URL.

Leave us a comment

logged inYou must be to post a comment.