A framework for linking literature-based knowledge integration and infrastructure-supported knowledge integration: Opportunities and challenges from a case study

1 Introduction

• Evidence synthesis faces fragmented and rapidly growing literature, methodological heterogeneity, and dependence on authors’ reporting choices, including how results and null findings are presented.Data, code, and computational workflows are also often insufficiently documented, limiting verifiability and reuse.Source: §1 · p. 3
• The framework distinguishes literature-based integration of reported results from infrastructure-supported integration that enables verification and reuse when datasets and workflows are accessible.The systematic review examines how far integration can move beyond publication-level evidence and where practical limits arise.Source: §1 · p. 4
• The case study reviews research on climate change, violent conflict, and household food security, focusing on knowledge-integration methods rather than the review’s substantive findings.The review’s topical findings are reported elsewhere (Awoke & Brück, 2026a).Source: §1 · p. 4
• Literature-based synthesis covered all 37 studies, whereas infrastructure-supported integration was feasible only for a small subset because reusable data and workflows were rarely accessible.Where outputs were available, synthesis could be produced directly from data and workflows rather than reconstructed from published findings.Source: §1 · pp. 4–5
• The paper’s contribution is a framework for understanding how the two approaches relate and can complement one another under different practical conditions.It connects evidence-synthesis research, open science and digital infrastructures, and sustainability research methodology.Source: §1 · p. 5

2.1 Data and knowledge

• Data are empirical observations, while knowledge is their interpretation and contextualization; access to underlying data and analytical processes enables claims to be verified and integrated.Integration may draw on reported claims or, when available, the data and workflows that account for them.Source: §2.1 · pp. 5–6

2.2 Knowledge integration

• The framework shifts attention to how research findings are represented and made accessible, because poor documentation and disciplinary formats hinder connection, comparison, and reuse across studies.It treats knowledge integration as depending partly on how research outputs are organized, accessed, and reusable.Source: §2.2 · pp. 7–8

integration

• The framework identifies coverage, structured synthesis, verifiability, and reusability as four dimensions for comparing knowledge integration across studies.These dimensions reflect dispersed evidence, heterogeneous methods and concepts, limited access to analytical procedures, and constraints on reusing datasets, code, and workflows.Source: integration · pp. 8–9
• Literature-based and infrastructure-supported integration are complementary: the former supports coverage and structured synthesis, while the latter can strengthen verifiability and reusability when outputs are accessible and documented.Infrastructure-supported integration links publications to underlying datasets, code, and analytical workflows.Source: integration · p. 9

Literature-based knowledge integration: Systematic literature review

• Systematic literature reviews provide structured procedures for identifying relevant research.Source: Literature-based knowledge… · p. 10
• Systematic literature reviews organize comparable published evidence across eligible studies, enabling pattern comparison and gap mapping regardless of data access, but reported analyses are difficult to verify or reuse.This approach synthesizes findings from studies even when their underlying datasets or analytical workflows are unavailable, but integration relies on reported results rather than direct access to analytical processes.Source: Literature-based knowledge… · pp. 10–11

Infrastructure-supported knowledge integration

• Infrastructure-supported integration connects publications to underlying research outputs when datasets, analytical code, and documentation are reusable; its feasibility depends on their availability and usability.The paper illustrates this approach with the TIB Knowledge Loom, an emerging open-science digital library of analysis-ready scientific knowledge.Source: Infrastructure-supported knowledge… · p. 11

Linking the two approaches

• Systematic reviews both synthesize reported findings and identify studies with reusable datasets and workflows, linking broad coverage to deeper infrastructure-supported integration where artefacts are available.The review workflow therefore serves both as a synthesis method and as a way to locate studies whose research outputs can support further integration.Source: Linking the two approaches · pp. 11–12

3.1 Case study context

• The case study examines whether infrastructure-supported integration can complement the literature-based synthesis, rather than revisiting the review’s substantive findings.It addresses relationships among climate change, violent conflict, and household food security across multiple disciplinary domains.Source: §3.1 · p. 12

Search engines for literature identification

• Searches in Scopus, Web of Science, and PubMed used combined terms for climate change, climate shocks, violent conflict, and household food security.Source: Search engines for literature… · p. 13

Reference management tools

• EndNote and Zotero organized bibliographic records, with Zotero enabling shared access to references.The tools supported documentation of search results and a transparent review workflow.Source: Reference management tools · p. 13

Screening and study selection

• PRISMA-guided selection combined duplicate removal, title-and-abstract screening, and full-text assessment, with Rayyan and ASReview supporting the workflow.ASReview prioritized potentially relevant records during title-and-abstract screening, with screening decisions documented transparently.Source: Screening and study selection · p. 13
• The review included 37 peer-reviewed studies after screening and full-text assessment.The case study focuses on quantitative survey studies of household food-security outcomes in relation to climatic and conflict-related stressors.Source: §3.1 · p. 12, Screening and study selection · p. 14

3.3 Assessing the feasibility of infrastructure-supported knowledge integration

• Feasibility assessment checked articles’ data-availability statements, public repositories, and author responses for datasets and analytical code.The assessment covered all 37 included studies.Source: §3.3 · p. 15

3.4 Operationalizing infrastructure-supported integration

• The study compared a manually coded knowledge gap map with a Loom-based workflow that reused structured data linked to studies’ findings, methods, code, and workflows.The review extracted variables including crisis type, food-security indicators, and reported relationship directions; authors reviewed Loom records and confirmed licensing arrangements before publication.Source: §3.4 · pp. 15–16

4.1 Data availability statements

• 54% of studies had no data-availability statement, 27% reported data available upon request, and 5% reported publicly available datasets and code.Another 5% reported datasets without accompanying code, and 5% stated that data were restricted or unavailable.Source: §4.1 · pp. 16–17, Table 1 · p. 17

4.2 Access to data and analytical workflows

• 81% of studies remained inaccessible because authors did not respond, and reported availability upon request rarely yielded usable data in practice.Among the studies assessed, reusable data and workflows came from 5% publicly available and 3% obtained upon request.Source: §4.2 · pp. 18–19
• Even when authors responded, data could be unavailable or conditional, and one study’s data and code could not be executed successfully.Source: §4.2 · p. 19

4.3 Outputs of infrastructure-supported knowledge integration

• The manually extracted knowledge gap map organized the evidence and highlighted areas with limited empirical research.Its construction depended on manual extraction and coding from publications.Source: §4.3 · p. 20, Figure 4 · p. 21
• Applying the same workflow to structured Loom data produced a partial knowledge gap map, demonstrating direct synthesis from reusable outputs rather than publication-level extraction.This example used only three studies, and broader application was constrained by limited reusable datasets and code.Source: §4.3 · p. 21, Figure 5 · p. 22

• Literature-based synthesis covered the review corpus, while infrastructure-supported integration extended synthesis depth where reusable outputs were available.The latter remained feasible only for a small subset because underlying data and workflows had to be accessible, documented, and reusable.Source: §5 · p. 23
• Publication-based synthesis supports broad integration, but it rarely enables direct engagement with the analytical processes behind findings.The discussion connects access to underlying data and workflows with verification of scientific findings.Source: §5 · pp. 23–24
• The Knowledge Loom linked findings to underlying data and workflows and enabled synthesis outputs to be generated directly from those materials.The case demonstrates this possibility under specific conditions, not broad feasibility across studies.Source: §5 · p. 24
• Extending integration beyond publications requires accessible datasets, executable code, clear documentation, and availability statements that reflect actual access.The discussion also identifies coordination among researchers, journals, funders, and infrastructures as relevant because reusability has institutional as well as technical conditions.Source: §5 · p. 24
• The study covers peer-reviewed, quantitative, English-language studies, excluding other evidence forms and potentially limiting generalizability.The small number of studies with reusable data and code also limits the empirical scope of infrastructure-supported integration; a larger sample may yield different results.Source: §5 · pp. 24–25

6 Conclusion

• Integration across studies remained largely limited to reported findings because reusable data and code were available for only a small subset, even where availability was indicated.The conclusion identifies research-output condition as shaping the scope of integration.Source: §6 · p. 25
• Literature-based integration remains essential for coverage and structured comparison across heterogeneous evidence, including identifying patterns and gaps through the manually built knowledge gap map.The map used information manually extracted from published articles.Source: §6 · p. 25
• Infrastructure-supported integration can generate synthesis directly from structured data and code rather than reconstructing it from publications, but this was feasible for only a small number of studies.Source: §6 · p. 26
• The approaches are complementary: literature-based integration supports coverage, while infrastructure-supported integration supports verifiability and reusability where outputs exist.Source: §6 · p. 26
• Advancing integration beyond reported findings depends on data and code being accessible, executable, and sufficiently documented for reuse.Source: §6 · p. 26