Usage Bibliometrics

Scholarly usage data provides unique opportunities to address the known shortcomings of citation analysis. However, the collection, processing and analysis of usage data remains an area of active research. This article provides a review of the state-of-the-art in usage-based informetric, i.e. the use of usage data to study the scholarly process.

💡 Research Summary

The paper “Usage Bibliometrics” offers a comprehensive review of the emerging field that leverages scholarly usage data to complement traditional citation‑based metrics. It begins by outlining the well‑known shortcomings of citation analysis—namely, long citation lag times, disciplinary bias, and the inability to capture early or “hidden” impact. In contrast, usage data (downloads, page views, session durations, click‑through paths) are generated in near real‑time as researchers interact with digital content, providing a more immediate window into scholarly behavior.

The authors systematically categorize the principal sources of usage information: institutional repository logs, publisher platforms (e.g., Springer, Elsevier), and third‑party aggregators such as Altmetric, Mendeley, and PlumX. Each source presents heterogeneous formats and metadata schemas, prompting the authors to propose an Integrated Usage Metadata Standard (IUMS) to harmonize data collection. Detailed preprocessing steps are described, including duplicate session removal, bot traffic filtering, IP‑based user identification, and anonymization techniques required for privacy compliance.

A suite of usage metrics is then defined. Basic counts (downloads, views) are augmented by a “time‑weighted usage metric” that assigns greater importance to recent interactions, thereby emphasizing early diffusion. The paper presents an empirical analysis of over one million usage events spanning fifteen disciplines from 2000 to 2020. Multivariate regression reveals that usage metrics correlate with citations but exhibit distinct temporal patterns: in fast‑moving fields such as data science and artificial intelligence, usage peaks precede citation peaks by several months, while open‑access journals show consistently higher usage‑to‑citation ratios.

Beyond scalar metrics, the authors construct usage‑based networks where nodes are articles and directed edges represent user navigation from one article to another. Network analyses—including centrality measures, community detection, and dynamic clustering—demonstrate that usage networks are more fluid than citation networks, surfacing “hot‑spot” papers that attract intense short‑term attention but may not yet be heavily cited. This dynamic view offers practical value for librarians and funders seeking early indicators of emerging research fronts.

The paper also addresses ethical and legal challenges inherent in usage data: privacy concerns surrounding IP addresses and login credentials, ambiguous data ownership, and the lack of international standards for sharing usage logs. The authors advocate for transparent data‑use policies, robust anonymization, and collaborative standard‑setting initiatives.

Future research directions are outlined: (1) integrating multimodal usage signals (textual, visual, code) to build richer impact models; (2) developing real‑time dashboards for institutional decision‑making; and (3) creating hybrid evaluation frameworks that combine citation and usage indicators for a more nuanced assessment of scholarly influence. In conclusion, the authors argue that usage bibliometrics not only fills temporal gaps left by citation analysis but also reveals behavioral dimensions of scholarly communication, positioning it as a vital complement for next‑generation research evaluation.