Approach

Our methodological approach is based on the following operational quality criteria:

  • Openness and transparency: Methodological assumptions are openly and clearly presented.
  • Coverage and accuracy: As detailed in graph.openaire.eu multiple data sources are ingested in the OpenAIRE research graph for coverage to the fullest extent possible, in order to provide meaningful indicators.
  • Clarity and replicability: We describe our construction methodology in detail, so that it can be verified and used by the scholarly communication community to create ongoing updates to our proposed statistics and indicators.
  • Readiness and timeliness: The methodology is built around well-established open databases and already tested knowledge extraction technologies - natural language processing (NLP)/machine-learning (ML) - using operational workflows in OpenAIRE to warrant timely results.
  • Trust and robustness: Our methodology also strives to be reliable, robust, and aligned to other assessment methods so that it can be operationalized, used and reused, in conjunction with other assessment methods.
The text above is modified from this report (DOI: 10.2777/268348).

Step-by-step

Starting from existing
research-related data sources
build an open, global
and trusted Research graph
we perform Statistical Analysis and produce Open Science Indicators
furthermore Network
Analysis
producing Collaboration Indicators
Often combine with external data (patents, social, company) and perform Impact Analysis to produce Innovation Indicators

Terminology and construction

  • Research Outcomes
    There are currently four different types of research outcomes in the OpenAIRE Research Graph:
    • publications
    • datasets
    • software
    • other research products
    OpenAIRE deduplicates (merges) different records of research outcomes and keeps the metadata of all instances.
    publication

    Research outcome intended for human reading (published articles, pre-prints, conference papers, presentations, technical reports, etc.)

    Only peer-reviewed publications are considered in the OS Observatory, unless explicitly stated otherwise (see tab Constructed Attributes for definition of peer-review).

    dataset

    Research data

    Granularity is not defined by OpenAIRE, it reflects the granularity supported by the sources from which the description of the dataset has been collected.

    software Source code or software package developed and/or used in a research context
    other research product Anything that does not fall in the previous categories (e.g. workflow, methods, protocols)
    More information at OpenAIRE Research Graph.