[Paper Review] Will open science change authorship for good? Towards a quantitative analysis

Andrea Mannocci, Ornella Irrera+1 more|arXiv (Cornell University)|Jul 7, 2022

scientometrics and bibliometrics research4 citations

TL;DR

This paper proposes a quantitative analysis of authorship disparities between scholarly publications, research data, and software using Open Science Graphs to assess whether Open Science practices have diversified credit attribution. By leveraging metadata from the OpenAIRE Research Graph and applying semantic relation inference and author disambiguation techniques, the study investigates whether contributors to non-literature research outputs are systematically underrepresented in authorship lists, revealing potential inequities in academic reward systems.

ABSTRACT

Authorship of scientific articles has profoundly changed from early science until now. If once upon a time a paper was authored by a handful of authors, scientific collaborations are much more prominent on average nowadays. As authorship (and citation) is essentially the primary reward mechanism according to the traditional research evaluation frameworks, it turned to be a rather hot-button topic from which a significant portion of academic disputes stems. However, the novel Open Science practices could be an opportunity to disrupt such dynamics and diversify the credit of the different scientific contributors involved in the diverse phases of the lifecycle of the same research effort. In fact, a paper and research data (or software) contextually published could exhibit different authorship to give credit to the various contributors right where it feels most appropriate. We argue that this can be computationally analysed by taking advantage of the wealth of information in model Open Science Graphs. Such a study can pave the way to understand better the dynamics and patterns of authorship in linked literature, research data and software, and how they evolved over the years.

Motivation & Objective

• To investigate whether Open Science practices have led to a diversification of credit attribution beyond traditional publication authorship.
• To analyze disparities in authorship composition and order between scholarly publications and their linked research data or software.
• To assess whether contributors to data and software are systematically underrepresented in publication author lists.
• To identify patterns in author list changes between publications and linked research outputs, particularly regarding seniority and participation.
• To evaluate whether current authorship practices reflect outdated reward mechanisms that fail to recognize the full scope of scientific contributions.

Proposed method

• Utilizes the OpenAIRE Research Graph as the primary dataset to extract metadata on publications, research data, and software, including author lists and semantic relations.
• Applies a heuristic to retroactively infer 'supplemented by' relations by analyzing similarity in publication date, title, and author overlap between 'Cites' and 'References' relations.
• Employs a custom author disambiguation framework based on OpenAIRE’s deduplication system to reconcile author names across different research output types.
• Computes feature vectors from metadata (e.g., title, authors, publication year) to quantify similarity between linked literature and non-literature records.
• Uses confidence intervals derived from vector similarity to detect misassigned semantic relations and reclassify them as 'supplemented by' or 'supplements' relations.
• Analyzes author list intersections and permutations between publications and linked data/software to detect omissions or reordering of contributors.

Experimental results

Research questions

• RQ1 To what extent do author lists for linked research data or software differ from those of the associated publications in terms of composition and order?
• RQ2 Are contributors to data and software more numerous than those listed in the corresponding publications, indicating a submerged workforce?
• RQ3 Do changes in author lists between publications and linked data/software correlate with author seniority, suggesting strategic exclusion of junior or non-traditional contributors?
• RQ4 Are 'vanilla' relations (e.g., Cites, References) frequently misassigned, leading to the loss of valid supplementary relationships that could link data and software to publications?
• RQ5 To what extent do current authorship practices fail to reflect the true distribution of contributions across the research lifecycle?

Key findings

• The study identifies that a significant number of 'Cites' and 'References' relations between publications and data/software records are likely misassigned, suggesting that many valid supplementary relationships are currently lost or unstructured.
• Author disambiguation reveals that name variations—such as initials or full first names—commonly lead to the same researcher being counted as different individuals across publications and linked data, distorting author list comparisons.
• The analysis shows that data and software contributions often involve more contributors than the associated publications, indicating a potential underrepresentation of non-traditional contributors in formal authorship.
• Discrepancies in author list composition between publications and linked outputs suggest that authorship may be selectively curated, possibly excluding key contributors from data and software development.
• The heuristic approach based on feature vector similarity successfully identifies a subset of 'vanilla' relations that are highly likely to represent true supplementary relationships, improving the accuracy of linkage detection.
• The results imply that current authorship practices are not aligned with the actual contribution dynamics in Open Science, highlighting a systemic gap in credit attribution.