利用Sankey圖呈現學術機構間論文發表量的比較：以某醫療資訊期刊的作者為例

Sankey Diagrams illustrating the comparison of publications by research institutes: An example for authors of a medical informatics journal

學術機構間論文發表量，經常利用各式圖表做分析及比較。唯未見採用桑基圖（Sankey diagram）來呈現其結果，尤其是針對資料的趨勢與轉折點的呈現。本研究蒐集醫療資訊雜誌2005至2021年論文作者的英文姓名（n=1，033），對照同名同姓於2017至2021年在PubMed發表論文的542位台灣作者，整理出7,013篇論文。利用社會網絡分析，集群出最多論文量的10個研究團隊代表，再比較研究團隊成員中的平均論文量。利用桑基圖呈現（1）六變項（如論文發表年度、作者城市、機構、期刊、論文類別、主題分類、及作者）的最多論文量排名、及（2）十個學術機構的年度論文量走勢。另以點圖（dot plot）呈現被引用量最多的100篇論文，橫軸為被引用量的百分位，縱軸為論文發表年度。研究結果顯示，（1）十個論文量最多的研究團隊，前三名的學者代表分別為林建羽、簡戊鑑、陳增基；（2）團隊成員平均論文數最多的學者代表為林建羽、簡戊鑑、陳得源，顯著高於其他七個團隊；（3）桑基圖容易呈現變項間的比較，與時間序列的熱點和資料趨勢；（4）點圖呈現被引用最多次（=2，417）的論文（PMID：30637094）係於2018年發表在J Extracell Vesicles期刊，作者之一是台大醫院胃腸肝膽科的主治醫師張毓廷。桑基圖比其他視覺圖更能呈現多變項資料間量的比較。本研究示範桑基圖的應用，提供未來利用視覺圖比較論文被引用量或學術成就的指標（如h指數），及其他領域數據分析之參考。

Several charts were used to compare the publications of research institutions. Sankey diagrams were not used by any of the studies to present their findings and results, especially when displaying trends and turning points in the reports. Based on articles published in the Journal of Taiwan Association for Medical Informatics since 2005, 1033 names of authors were matched with authors of 7013 articles indexed in PubMed from 2017 to 2021. We used cluster analysis to identify the top ten research teams using social network analysis. The mean number of publications among the teams was compared using a forest plot. A Sankey diagram was drawn to illustrate the trends of publications in ten research institutes based on six variables (year, city, institute, journal, document type, subject category, and author). We constructed a dot plot of the top 100 most cited articles using the percentiles of citations and publication years as axes x and y, respectively. We observed that (1) three of the top research teams were represented by Chien-Yu Lin, Wu-Chien Chien, and Tzeng-Ji Chen; (2) more mean publications in research teams have been authored by Chien-Yu Lin, Wu-Chien Chien, and Der-Yuan Chen; (3) the perception of one look worthy of 1000 words was evident in Sankey when compared to other types of visualizations; and (4) the most cited article (with a PMID of 30637094 and 2417 citing articles) in the dot plot was published in J Extracell Vesicles in 2018 and was written by Yu-Ting Chang, an attending physician at Taiwan University Hospital who is also an attending physician of the Hepatobiliary Department. Sankey's visualization is superior to other visualizations, particularly when used to highlight a number of dimensions within a graph with a glance. Sankey's visualizations were demonstrated and recommended for applications in bibliometrics and other areas of data analysis.