Human-In-The-Loop Artificial Intelligence for Account Reconciliation and Financial-Reporting Transparency: A Bibliometric and Science-Mapping Analysis

Abstract

Human-in-the-loop artificial intelligence (AI) is increasingly examined in accounting workflows involving transaction matching, exception detection, and financial close. This study aimed to map research connecting AI-assisted account reconciliation, human oversight, and financial-reporting transparency through bibliometric and science-mapping methods. The proposed search covered Scopus and Web of Science publications from 2015 to 2025. Records were screened for relevance, and bibliographic data were prepared for publication, citation, collaboration, and keyword analyses. In the screening example, 1,248 records were retrieved, 286 duplicates were removed, and 962 records were screened by title and abstract. After 261 full texts were assessed, 44 publications were included. The example corpus comprised 30 journal articles, eight conference papers, and six reviews; 36 publications, or 81.8%, were multi-authored. The dataset accumulated 612 citations, with a median of eight citations per publication. A count model estimated annual publication growth at an incidence rate ratio of 1.20, with a 95% confidence interval of 1.08 to 1.34 and a p-value of .001. Network mapping produced five author co-authorship clusters and four reference co-citation clusters. Keyword analysis retained 24 terms and identified five thematic groups addressing accounting automation, reconciliation and financial close, audit analytics and internal control, human oversight and explainability, and transparency and reporting quality. The comparison of citation distributions by document type was statistically significant before adjustment; after citation-age adjustment, it was not significant (p = .086). These example results demonstrate how bibliometric indicators can describe publication activity, citation patterns, collaboration, and conceptual organization. They do not establish that AI improves reconciliation accuracy, human decisions, or financial-reporting outcomes. Final conclusions require verified database records, screening decisions, citation data, and network analyses. The study’s intended contribution is an integrated quantitative map of research at the intersection of AI-supported reconciliation, human review, and transparent financial reporting.