图灵奖得主识别与预测研究——基于多文献计量指标和支持向量机

Study on the Identifiability and Predictability of Turing Award Winners Based on Multiple Bibliometric Indicators and Support Vector Machine

国内外学者开展了若干借助文献计量指标来识别和预测重要科技奖项得主的研究与实践,但已有研究大多局限于对少数几项文献计量指标进行简单的计量统计,对问题的揭示不够全面和深入。利用支持向量机对图灵奖得主和非图灵奖得主的多项文献计量指标进行了分析,在两种不同情境下借助支持向量机对样本数据进行分类学习并进行识别与预测,发现利用文献计量指标建立的支持向量机模型对图灵奖得主具有很好的识别能力,但预测能力一般。

Researchers have explored the use of bibliometric indicators to identify and predict the winners of some prominent science a- wards. Previous work mainly employed some preliminary bibliometrics and statistics, lacking coverage of bibliometric indicators and depth of analysis. In order to investigate the identifiability and predictability of Turing Award wimaers, 20 bibliometric indicators about 33 Turing Award winners and 300 non-Turing Award winners were analyzed. A classification and prediction analysis was conducted with the biblio- metric indicators and data by support vector machine（SVM） in two different scenarios. The result indicated that a classification model could be developed based on bibilometric indicators to identify the Turing Award winners with a high precision. However, the prediction precision appeared medium.