科学计量学研究中人工智能算法提及情况与演化趋势分析OACHSSCD

[Purpose/Significance]As intelligent digital technologies increasingly integrate into scientific research,research paradigms and analytical methods in scientometrics continue to evolve.The study aims to reveal the mention pa-tterns and evolutionary trends of AI algorithms in scientometrics literature.[Method/Process]Based on 6,989 full-text articles published in Scientometric since its inception,this study conducted full-text data extraction,applyed deep learning models for automatic section classification,and employed a fine-tuned large language model to automatically extract AI algo-rithm entities.[Result/Conclusion]Research indicates that AI and non-AI algorithms exhibit a clear pattern of functional differentiation and complementarity within scientometrics.Non-AI algorithms are characterized by broad coverage and stable occurrence,serving as foundational tools embedded throughout the research workflow for basic analysis and robust-ness checks.In contrast,AI algorithms demonstrate higher discussion intensity within individual publications,with con-centrated use in methodological design and results analysis,where they undertake core tasks such as complex pattern recog-nition,semantic modeling,and predictive inference.In recent years,AI models represented by BERT and GPT have expe-rienced rapid growth,and some of these algorithms simultaneously achieve high coverage and high levels of attention,driving a methodological shift from traditional statistical analysis toward text semantic understanding and complex system modeling.AI algorithms play a key role in enhancing analytical capabilities and driving methodological innovation in scientometrics.