利用R和新演算法視覺化分析學術期刊的作者合作情形

Cluster Analysis of Author Collaborations in Scholarly Journals Using R and a Novel Algorithm

本研究提出一種新穎演算法，稱為主從關係演算法(follower-leader clustering, FLC)。藉由統計軟體R，可有效地聚類分析和視覺化展示學術期刊中的作者合作。該演算法採用分層聚類方法，根據作者與潛在領導者的主要聯繫和他們在領導者活動中的參與程度將作者分類。然後藉由視覺化展示清晰簡明的集群關係，使讀者能夠輕鬆識別作者合作的情形。為評估所提出演算法的有效性，我們將其應用於《醫療資訊雜誌》的大型學術論文資料。結果顯示，該演算法可以有效地將前20論文量最多的作者，區分出六個集群，並提供多樣式的視覺圖呈現分類的結果。本研究利用主從關係演算法，聚類分析作者合作。再用R語言視覺化展示集群的結果。所提出的演算法為研究人員探索作者合作的分類，並可作為未來共字分析的分類參考。

This paper proposes a novel algorithm for analyzing author collaborations in scholarly journals using cluster analysis and visual displays. The algorithm called follower-leader clustering (FLC) is implemented using the statistical software R, which allows for efficient and reproducible analysis of large datasets. It utilizes a hierarchical clustering approach to group authors according to (1) their principal connection to the potential leader and (2) their involvement in the leader's activities. Visual displays are then used to represent these groups in a clear and concise manner, allowing readers to easily identify patterns in author collaborations. To evaluate the effectiveness of the proposed algorithm, we applied it to a large dataset of scholarly articles from The Journal of Taiwan Association for Medical Informatics (JTAMI). The results show that the algorithm can effectively identify clusters of authors into six clusters, as well as provide valuable insights into the structure of author networks within a given field. We demonstrate the potential of using cluster analysis and visual displays for analyzing author collaborations in scholarly journals. With the proposed algorithm, researchers are able to explore and understand the complex relationships between authors and cowords or cooccurrences, and can easily apply it to a wide range of datasets and fields of study.