| 李勃慧,王运红,杨代庆,郑楚华,陈国娇.基于机器学习的科技人才综合学术水平评价
指标与模型研究[J].情报工程,2024,10(5):115-127 |
| 基于机器学习的科技人才综合学术水平评价
指标与模型研究 |
| Research on Comprehensive Academic Level Evaluation Indicators and Models for Scientific and Technological Talents Based on Machine Learning |
| |
| DOI：10.3772/j.issn.2095-915X.2024.05.010 |
| 中文关键词: 科技人才；人才评价；人才识别；评价模型 |
| 英文关键词: Scientific and Technological Talents; Talent Evaluation; Talent Identification; Evaluation Models |
| 基金项目:国家社会科学基金项目“大数据环境下同行评议方法模式研究”（19BTQ082）。 |
| 作者 | 单位 | | 李勃慧 | 中国科学技术信息研究所 北京 100038 | | 王运红 | 中国科学技术信息研究所 北京 100038 | | 杨代庆 | 中国科学技术信息研究所 北京 100038 | | 郑楚华 | 中国科学技术信息研究所 北京 100038 | | 陈国娇 | 中国科学技术信息研究所 北京 100038 |
|
| 摘要点击次数: 42 |
| 全文下载次数: 18 |
| 中文摘要: |
| [目的/意义]从学术诚信、科研产出、科研活动和学术声誉4个维度，打破单一指标局限性，为人才评价中“破四唯”“立新标”提供新的思路和参考。[方法/过程]针对科技人才的特征，构建了98项定量指标与定性特征指标。选取研究样本，进行数据集构建和数据预处理，形成最终识别预测模型的输入数据；通过机器学习获得样本特征指标与评价结果之间的隐含关系，比较9种模型算法下的人才学术能力表现，综合得出最优模型进行后续训练和调优，通过对比模型输出与实际结果进行模型评价。[局限]模型仅基于实验样本数据量和特征，推广应用还需要更多的样本数据进行训练调优。[结果/结论]实证研究表明，模型对高水平科技人才评价具备较好的适用性，为科技人才评价提供新的视角和方法。 |
| 英文摘要: |
| [Objective/Significance] This study breaks the limitations of a single indicator from four dimensions: academic
integrity, research output, research activities, and academic reputation, providing new ideas and references for breaking the “4
Single dimension” and “setting new standards” in talent evaluation. [Methods/Processes] For the characteristics of scientific and
technological talents, 98 quantitative indicators and qualitative characteristic indicators were constructed. This study selected
research samples for dataset construction and data preprocessing to form the input data for the final recognition and prediction model; Obtain the implicit relationship between sample feature indicators and evaluation results through machine learning,
compare the academic performance of talents under 9 model algorithms, comprehensively determine the optimal model for
subsequent training and optimization, and evaluate the model by comparing the model output with actual results. [Limitations]
This research model is only based on experimental sample data size and features, and further training and optimization of sample
data are needed for its widespread application. [Results/Conclusions] Empirical research has shown that the model has good
applicability for the evaluation of scientific and technological talents, providing new perspectives and methods for the evaluation
of scientific and technological talents. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
