2.6 Citation Genealogy Analysis for Cultural Heritage Research – Tool Development

1. Background and project aim

The Citation Genealogy Analysis method

The Citation Genealogy Analysis tool is based on a method for tracing how scholarly statements develop, are replicated, and can transform through unreflected citation and re-citation of contestable information in historical publications and now digital systems. The method utilises systematic network analysis approaches in combination with established humanities research processes to address how replication patterns can be fundamental aspects of understanding citation genealogy and research foundations.

The tool offers a systematic and low-threshold option for reviewing historical origins and development of contestable data – particularly valuable for humanities researchers with limited technical training.

The approach was initially outlined during participation in the Wikimedia Germany Open Science Fellows Program (2020–2021) and significantly expanded during the developer's MSc dissertation in Digital Scholarship at the University of Oxford (2022–2023).

Why a web tool?

The Citation Genealogy Analysis method demonstrated significant value as an expansive complementary research approach, but remained difficult to apply without technical expertise or network theory knowledge. Creating a digital tool removes this barrier by making the method accessible, transparent, and reusable for humanities researchers and students without programming experience, while preserving the methodological value of citation network analysis.

Current scope: publication-level network visualisation

The Citation Genealogy Web Tool, in its current form, expands from existing citation management software such as Zotero and Citavi to enable humanities researchers to visualise their personal literature collection as a network graph, identifying patterns in how their research corpus is structured. This is particularly helpful for research fields relying on gathering and working from a state of the literature for their research question. Humanities researchers may not be accustomed to seeing research as an interconnected field shaped by replication patterns and citation network mechanisms. The tool provides a widening approach to exploring and understanding research by revealing these underlying structures.

The tool currently operates at the title level and requires prior citation linking in Zotero or a chosen citation management software – with the tool itself subsequently visualising which publications cite which others, revealing clusters, showing influence patterns, and - crucially - temporal evolution. This network visualisation reveals patterns invisible in raw literature lists or bibliographic datasets.

2. What has been built

Foundation and technical architecture

Development began with refactoring of the original dissertation code to establish a stable foundation and expansion for the web tool. The prototype is implemented in Python and leverages open-source libraries, including pandas for data handling, networkx for citation graphs, matplotlib/PyVis/Jaal for visualisations, and Streamlit for the web interface. Python was chosen for flexible future expansion, and since its libraries provide approachable and aesthetic options for humanities-focused usability.

The tool's analytical workflow

The tool implements five stages:
(1) import bibliographic metadata from CSV (with expansion for BibTeX and RIS to be implemented)
(2) clean and normalise data
(3) generate edgelist and nodelist
(4) calculate network metrics (such as in-degree, out-degree, betweenness centrality, closeness centrality)
(5) visualise through multiple interactive approaches for engaging and insightful exploration

Fig. 1: Start page of the Citation Genealogy Analysis tool with overview and upload function.

Current features

Tab 1: Data overview with key metrics, publication timeline with statistical markers, most influential publications table, and publication type distribution.

Fig. 2: After upload, the data overview currently shows key metrics, a publication timeline with statistical markers, the most influential publications table, and publication type distribution. This selection can be adapted based on trial feedback.

Tabs 2–3: Genealogy visualisation with hierarchical layout by year, colour-coded by reference type, node sizing by citation frequency, with hover details and network metrics. The dynamic filtering by year range, publication type, minimum citations, and language provides especially engaging and flexible functions.

Fig. 3: The central exploration tabs of the tool provide multiple approaches to network graphs, focusing on hierarchical visualisation in chronological order to allow scholars and students to understand their data as historically grown organisms with distinctive patterns that help us understand the evolution of the research better.

Tab 4: Enhanced analysis including author statistics, language distribution, detailed timeline with percentile markers, and data quality assessment with missing citation identification.

Fig. 4: A helpful zooming-in dynamic helps to section off specific pattern frames and understand details of citation genealogy. The visualisations here are text-free for visual focus, but descriptions can be added via the optional toggles in the tool.

Tab 5: Pattern detection identifying citation cycles and longest influence chains with visual highlighting, offering additional insightful visualisation options.

Tab 6: Interactive networks via Jaal and PyVis, plus data export options (edgelists, nodelists, patterns, complete datasets).

Fig. 5: The tool also incorporates alternative interactive networks based on the existing Jaal and PyVis libraries to allow for dynamic exploration of the data in a non-chronological form.

3. Current status and next steps

Operational prototype

A functioning prototype has been established and tested on the case study dataset. Researchers can import data, construct networks, calculate metrics, detect patterns, and export results.

Planned Interface improvements include a label toggle for node labels, improved vertical spacing on hierarchical layouts, add options to include/exclude isolated nodes, and implement scrollable containers for dense networks. Furthermore, visualisation enhancements to integrate additional visualisation control panels, add legends and variant format support, including BibTeX, and RIS import.

Structured review and enhancement

The review phase will include structured community outreach and feedback from technical specialist colleagues from the Centre for Digital Scholarship (Oxford) and community contacts working on citation integrations for Zotero, such as Diego de la Hera (Cita), and Tim Wölfle (Local Citation Network), to further review visualisation design, and strategic alignment with existing infrastructure.

In addition, the tool will be tested with library professionals (University of Heidelberg Library), who will provide feedback on integration with institutional systems, discoverability, and documentation requirements. It will also be tested with target audience end users – including humanities students in Heidelberg and Berlin – to gather pedagogical feedback on usability, visual design clarity, and relevance to actual research practice.

Following review, tool release is planned through dual channels: GitHub for open-source accessibility and community contribution, and Heidelberg University Library for institutional stewardship and NFDI discoverability.

Long-term development: statement-level analysis

Statement-level analysis capabilities are planned and will provide a significant tool enhancement, which will incorporate a central function of the Citation Genealogy Analysis method. This will enable manual enrichment of citation data with text excerpts of relevant citations or central statements, research questions, relevant custom categorical tags, and metadata on citation nature (agreement, disagreement, modification, contestation). This enrichment will move beyond publication-level patterns to examine how individual arguments develop and circulate through scholarly discourse and will transform the tool from publication-level network analysis to statement-level genealogy tracing, realising the methodological potential of the method.

4. Contribution to NFDI and conclusion

The tool contributes to NFDI objectives through FAIR research data practices (structured export formats), methodological transparency (making visible how citation networks structure research), and infrastructure tailored to humanities workflows (accommodating historical sources, incomplete data, and interpretive research practices).

As an open-source resource designed for iterative enhancement and published through institutional channels, the Citation Genealogy Analysis Web Tool is positioned for ongoing community contribution, integration with NFDI infrastructure, and application across humanities disciplines. By making visible the persistence of contestable information and the replication patterns that shape scholarly discourse, the tool supports the kind of critical, reflexive research practice that humanities scholarship demands.

Charlotte Oertel