The extension of zbMATH Open by arXiv preprints

TL;DR

The paper describes extending zbMATH Open with unpublished arXiv preprints to expand coverage and visibility for mathematical work. It implements a two-step, DOI-informed matching strategy followed by a machine-learning-based classifier to link arXiv preprints to zbMATH Open records, while distinguishing them from peer-reviewed publications. It also updates author disambiguation to accommodate arXiv-based entries and presents a subset-based scope decision to balance coverage with quality. The initiative increases the Open Access fraction, provides richer searchability (including abstracts), and opens avenues for future integrations (software data, interlinks, and full-text capabilities) via open APIs, encouraging community involvement.

Abstract

zbMATH Open has started a new feature -- relevant preprints posted at arXiv will also be displayed in the database. In this article we introduce this new feature and the underlying editorial policy. We also describe some of the technical issues involved and discuss the challenges this presents for future developments.

Figures (5)

• Figure 1: New filter functions allow to clearly distinguish additional unpublished arXiv entries
• Figure 2: Filter functions for more than 200,000 newly added arXiv preprints
• Figure 3: zbMATH author profiles with additional arXiv preprints
• Figure 4: A withdrawn arXiv submission with a separator error
• Figure 5: A new author profile based on arXiv submissions