Survey of 100 scientometrics and research-evaluation respondents in Tehran finds new-technology use explains 73% of variance in evaluation accuracy, 64% in fairness, 70.2% in transparency, and 61.1% in responsiveness

The study proposes and tests a positive association between new-technology use and the quality of scientometric governance, operationalized through four governance indicators: accuracy, fairness, transparency, and responsiveness. Prior discussion has largely stayed with the limitations of traditional indicators (citation counts, the h-index, journal impact factors), such as author name ambiguities, indicator instability, linguistic and regional biases, and difficulties in identifying emerging trends; this study links those challenges to the governance role of machine learning, data mining, natural language processing, dynamic network analysis, and artificial intelligence, and offers a testable quantitative association. Based on a questionnaire survey of 100 respondents in Tehran, using a researcher-designed Likert scale and simple regression in SPSS 26; reported coefficients of determination are 73% for accuracy, 64% for fairness, 70.2% for transparency, and 61.1% for responsiveness.

The study defines scientometric governance as a set of structures, rules, processes, and tools that determine "what and how to measure," and argues that technology reshapes this governance arrangement. The author integrates new technologies with scientific governance theories, proposing that technology moves scientometric governance from a centralized, opaque process toward an open, verifiable, data-consensus-based system and shifts accountability from a moral obligation to a structural imperative. This is a theoretical integration and concluding argument that is mutually supportive with the regression results; the text does not report an independent test of the theoretical framework.

For policy and applied settings, the study argues that if indicators lack sufficient accuracy or transparency, macro-level scientific policies may lead to incorrect or unfair decisions, so technology adoption can promote researchers' trust, reduce errors, and improve evaluation efficiency. It directly connects technology adoption to evidence-based policymaking and to more adaptive and sustainable scientific evaluation systems, emphasizing that data quality, model validation, and auditing require clear standards. Derived from the author's inferences and conclusion based on the survey results; the text reports no policy-level intervention experiment or long-term tracking data.

The study addresses policymakers in research evaluation and scientometric governance, managers of university research-evaluation units, and providers of scientific databases and scientometric services, and it applies to evaluation settings that rely on indicators to support resource allocation and science strategy design. Its conclusions point to introducing machine learning, data mining, natural language processing, dynamic network analysis, and artificial intelligence into the scientometric process to improve accuracy, fairness, transparency, and responsiveness, and to establishing standards for data quality, model validation, and auditing. Because the loaded text is incomplete and lacks figures, questionnaire items, and full references, readers who want to design a concrete evaluation workflow should return to the original to verify the measurement instrument and variable definitions.

Readers should still watch: the limits of convenience sampling and a 100-person sample for population inference; that simple regression shows association rather than causation; that accuracy, fairness, transparency, and responsiveness are all measured by a researcher-designed Likert scale whose construct validity awaits further study; and that it is unclear whether the high coefficients of determination are affected by common-method bias. In addition, the loaded text is incomplete and lacks figures and full references, so questionnaire items, variable definitions, and model specification cannot be verified—open questions to confirm when reading the original.